Properties of concrete

Fifth Edition
By

Otieno O. Christopher

Current CEO at Otihome design

Structural engineer

Former student at Technical University Of Mombasa

Email: christopherooko61@gmail.com

1
Properties of concrete ............................................................................................................................... 1
Chapter 1. Portland cement ............................................................................................................. 7
Historical note .......................................................................................................................... 7
Manufacture of Portland cement .............................................................................................. 8
Chemical composition of Portland cement ............................................................................. 13
Hydration of cement ............................................................................................................... 18
Setting ..................................................................................................................................... 24
Fineness of cement ................................................................................................................. 25
Structure of hydrated cement .................................................................................................. 31
Volume of products of hydration ............................................................................................ 31
Mechanical strength of cement gel ......................................................................................... 41
Water held in hydrated cement paste ...................................................................................... 42
Heat of hydration of cement ................................................................................................... 43
Influence of the compound composition on properties of cement .......................................... 48
Tests on properties of cement ................................................................................................. 56
References .............................................................................................................................. 64
Chapter 2. Cementitious materials of different types .................................................................... 70
Categorization of cementitious materials* .............................................................................. 70
Different cements ................................................................................................................... 73
Ordinary Portland cement ....................................................................................................... 78
Rapid-hardening Portland cement .......................................................................................... 79
Special very rapid-hardening Portland cements ..................................................................... 80
Low heat Portland cement ...................................................................................................... 83
Sulfate-resisting cement ......................................................................................................... 84
White cement and pigments.................................................................................................... 85
Portland blastfurnace cement .................................................................................................. 87
Supersulfated cement .............................................................................................................. 90
Pozzolanas .............................................................................................................................. 91
Silica fume .............................................................................................................................. 94
Fillers ...................................................................................................................................... 95
Other cements ......................................................................................................................... 96
Which cement to use ............................................................................................................... 97
High-alumina cement ............................................................................................................. 98
Conversion of high-alumina cement ..................................................................................... 102
Refractory properties of high-alumina cement ..................................................................... 110
References ............................................................................................................................ 111
Chapter 3. Properties of aggregate .............................................................................................. 117
General classification of aggregates ..................................................................................... 117
Classification of natural aggregates ...................................................................................... 118
Sampling ............................................................................................................................... 120
Particle shape and texture ..................................................................................................... 122
Bond of aggregate ................................................................................................................. 128
Strength of aggregate ............................................................................................................ 129
Other mechanical properties of aggregate ............................................................................ 133
Specific gravity ..................................................................................................................... 136
Bulk density .......................................................................................................................... 137
Porosity and absorption of aggregate.................................................................................... 138
Moisture content of aggregate .............................................................................................. 141
Bulking of fine aggregate ..................................................................................................... 143
Deleterious substances in aggregate ..................................................................................... 145
Soundness of aggregate ........................................................................................................ 150
Alkali–silica reaction ............................................................................................................ 152
Alkali–carbonate reaction ..................................................................................................... 156
Thermal properties of aggregate ........................................................................................... 157
Sieve analysis ....................................................................................................................... 158
Grading requirements ........................................................................................................... 166
Practical gradings ................................................................................................................. 173

2
 Grading of fine and coarse aggregates .................................................................................. 177
Gap-graded aggregate ........................................................................................................... 182
Maximum aggregate size ...................................................................................................... 185
Use of ‘plums’ ...................................................................................................................... 186
Handling of aggregate .......................................................................................................... 187
Special aggregates ................................................................................................................ 187
Recycled concrete aggregate ................................................................................................ 188
Chapter 4. Fresh concrete ............................................................................................................ 190
Quality of mixing water ........................................................................................................ 190
Density of fresh concrete ...................................................................................................... 192
Definition of workability ...................................................................................................... 193
The need for sufficient workability ...................................................................................... 194
Factors affecting workability ................................................................................................ 194
Measurement of workability ................................................................................................. 197
Comparison of tests .............................................................................................................. 207
Stiffening time of concrete ................................................................................................... 212
Effect of time and temperature on workability ..................................................................... 212
Segregation ........................................................................................................................... 215
Bleeding ................................................................................................................................ 217
The mixing of concrete ......................................................................................................... 218
Ready-mixed concrete .......................................................................................................... 226
Retempering ......................................................................................................................... 227
Pumped concrete................................................................................................................... 228
Shotcrete ............................................................................................................................... 235
Underwater concrete ............................................................................................................. 237
Preplaced aggregate concrete ............................................................................................... 238
Vibration of concrete ............................................................................................................ 240
Revibration ........................................................................................................................... 243
Vacuum-dewatered concrete ................................................................................................ 244
Analysis of fresh concrete .................................................................................................... 246
Self-compacting (self-consolidating) concrete ..................................................................... 247
References ............................................................................................................................ 248
Chapter 5. Admixtures .................................................................................................................. 256
Benefits of admixtures .......................................................................................................... 256
Types of admixtures ............................................................................................................. 256
Accelerating admixtures ....................................................................................................... 257
Retarding admixtures ............................................................................................................ 262
Water-reducing admixtures .................................................................................................. 266
Superplasticizers ................................................................................................................... 269
Special admixtures ................................................................................................................ 277
Remarks about the use of admixtures ................................................................................... 279
References ............................................................................................................................ 280
Chapter 6. Strength of concrete .................................................................................................... 284
Water/cement ratio ............................................................................................................... 284
Effective water in the mix..................................................................................................... 289
Gel/space ratio ...................................................................................................................... 290
Porosity ................................................................................................................................. 293
Influence of properties of coarse aggregate on strength ....................................................... 301
Influence of aggregate/cement ratio on strength ................................................................... 305
Nature of strength of concrete .............................................................................................. 308
Microcracking....................................................................................................................... 318
Aggregate–cement paste interface ........................................................................................ 321
Effect of age on strength of concrete .................................................................................... 322
Maturity of concrete ............................................................................................................. 325
Relation between compressive and tensile strengths ............................................................ 331
Bond between concrete and reinforcement ........................................................................... 333
References ............................................................................................................................ 334

3
 Chapter 7. Further aspects of hardened concrete ....................................................................... 342
Curing of concrete ................................................................................................................ 342
Autogenous healing .............................................................................................................. 354
Variability of strength of cement .......................................................................................... 355
Changes in the properties of cement ..................................................................................... 358
Fatigue strength of concrete.................................................................................................. 362
Impact strength ..................................................................................................................... 371
Electrical properties of concrete ........................................................................................... 375
Acoustic properties ............................................................................................................... 381
References ............................................................................................................................ 384
Chapter 8. Temperature effects in concrete ................................................................................ 389
Influence of early temperature on strength of concrete ........................................................ 389
Steam curing at atmospheric pressure .................................................................................. 396
High-pressure steam curing (autoclaving) ............................................................................ 401
Other thermal curing methods .............................................................................................. 404
Thermal properties of concrete ............................................................................................. 405
Coefficient of thermal expansion .......................................................................................... 408
Strength of concrete at high temperatures and resistance to fire .......................................... 417
Strength of concrete at very low temperatures ..................................................................... 424
Mass concrete ....................................................................................................................... 427
Concreting in hot weather ..................................................................................................... 432
Concreting in cold weather ................................................................................................... 434
References ............................................................................................................................ 439
Chapter 9. Elasticity, shrinkage, and creep................................................................................... 447
Stress–strain relation and modulus of elasticity .................................................................... 447
Expressions for modulus of elasticity ................................................................................... 453
Dynamic modulus of elasticity ............................................................................................. 455
Poisson’s ratio ...................................................................................................................... 456
Early volume changes ........................................................................................................... 458
Autogenous shrinkage .......................................................................................................... 460
Swelling ................................................................................................................................ 461
Drying shrinkage .................................................................................................................. 461
Factors influencing shrinkage ............................................................................................... 465
Prediction of shrinkage ......................................................................................................... 474
Differential shrinkage ........................................................................................................... 475
Shrinkage-induced cracking ................................................................................................. 478
Moisture movement .............................................................................................................. 480
Carbonation shrinkage .......................................................................................................... 481
Shrinkage compensation by the use of expansive cements* ................................................. 484
Creep of concrete* ................................................................................................................ 487
Factors influencing creep ...................................................................................................... 491
Relation between creep and time .......................................................................................... 507
Nature of creep ..................................................................................................................... 511
Effects of creep ..................................................................................................................... 514
References ............................................................................................................................ 516
Chapter 10. Durability of concrete ............................................................................................... 526
Causes of inadequate durability ............................................................................................ 526
Transport of fluids in concrete .............................................................................................. 527
Diffusion ............................................................................................................................... 529
Absorption ............................................................................................................................ 530
Water permeability of concrete ............................................................................................ 533
Air and vapour permeability ................................................................................................. 539
Carbonation .......................................................................................................................... 542
Acid attack on concrete ........................................................................................................ 550
Sulfate attack on concrete ..................................................................................................... 552
Efflorescence ........................................................................................................................ 558
Effects of sea water on concrete ........................................................................................... 559

4
 Disruption by alkali–silica reaction ...................................................................................... 562
Abrasion of concrete ............................................................................................................. 566
Erosion resistance ................................................................................................................. 568
Cavitation resistance ............................................................................................................. 569
Types of cracking ................................................................................................................. 570
References ............................................................................................................................ 574
Chapter 11. Effects of freezing and thawing and of chlorides ...................................................... 583
Action of frost ....................................................................................................................... 583
Air entrainment ..................................................................................................................... 592
Entrained-air requirements ................................................................................................... 595
Tests of resistance of concrete to freezing and thawing ....................................................... 604
Further effects of air entrainment ......................................................................................... 607
Effects of de-icing agents ..................................................................................................... 609
Chloride attack* .................................................................................................................... 611
Chlorides in the mix ............................................................................................................. 615
Ingress of chlorides ............................................................................................................... 615
Threshold content of chloride ions ....................................................................................... 617
Influence of blended cements on corrosion .......................................................................... 619
Further factors influencing corrosion ................................................................................... 620
Tests for penetrability of concrete to chlorides ..................................................................... 623
Stopping corrosion ................................................................................................................ 623
References ............................................................................................................................ 624
Chapter 12. Testing of hardened concrete ................................................................................... 631
Tests for strength in compression ......................................................................................... 631
Effect of end condition of specimen and capping ................................................................. 634
Testing of compression specimens ....................................................................................... 637
Failure of compression specimens ........................................................................................ 639
Effect of height/diameter ratio on strength of cylinders ....................................................... 640
Comparison of strengths of cubes and cylinders .................................................................. 643
Tests for strength in tension .................................................................................................. 645
Influence on strength of moisture condition during test ....................................................... 651
Influence of size of specimen on strength ............................................................................ 652
Test cores .............................................................................................................................. 663
Cast-in-place cylinder test .................................................................................................... 670
Influence of rate of application of load on strength .............................................................. 671
Accelerated-curing test ......................................................................................................... 673
Non-destructive tests ............................................................................................................ 677
Rebound hammer test ........................................................................................................... 677
Penetration resistance test ..................................................................................................... 681
Pull-out test ........................................................................................................................... 682
Post-installed tests ................................................................................................................ 684
Ultrasonic pulse velocity test ................................................................................................ 685
Further possibilities in non-destructive testing ..................................................................... 687
Resonant frequency method ................................................................................................. 688
Tests on the composition of hardened concrete .................................................................... 689
Variability of test results ....................................................................................................... 691
References ............................................................................................................................ 697
Chapter 13. Concretes with particular properties ........................................................................ 706
Concretes with different cementitious materials................................................................... 706
Concrete containing fly ash .................................................................................................. 709
Concretes containing ground granulated blastfurnace slag (ggbs) ....................................... 718
Concrete containing silica fume ........................................................................................... 722
High performance concrete................................................................................................... 731
Properties of aggregate in high performance concrete.......................................................... 733
Aspects of high performance concrete in the fresh state ...................................................... 734
Aspects of hardened high performance concrete .................................................................. 738
Chapter 14. Selection of concrete mix proportions (mix design) ................................................. 744

5
 Cost considerations ............................................................................................................... 744
Specifications ....................................................................................................................... 745
The process of mix selection ................................................................................................ 747
Mean strength and ‘minimum’ strength ................................................................................ 748
Quality control ...................................................................................................................... 758
Factors governing the selection of mix proportions ............................................................. 759
Mix proportions and quantities per batch ............................................................................. 766
Combining aggregates to obtain a type grading ................................................................... 768
x:y:z = 1:0.94:2.59 ................................................................................................................................ 769
American method of selection of mix proportions ............................................................... 772
Mix selection for high performance concrete ....................................................................... 778
Mix selection for lightweight aggregate concrete ................................................................. 779
British method of mix selection (mix design)....................................................................... 783
Other methods of mix selection ............................................................................................ 791
Concluding remarks .............................................................................................................. 792
References .................................................................................................................................... 792

6
Chapter 1. Portland cement

Cement, in the general sense of the word, can be described as a material with adhesive
and cohesive properties which make it capable of bonding mineral fragments into a
compact whole. This definition embraces a large variety of cementing materials.
For constructional purposes, the meaning of the term ‘cement’ is restricted to the
bonding materials used with stones, sand, bricks, building blocks, etc. The principal
constituents of this type of cement are compounds of lime, so that in building and
civil engineering we are concerned with calcareous cement. The cements of interest in
the making of concrete have the property of setting and hardening under water by
virtue of a chemical reaction with it and are, therefore, called hydraulic cements.
Hydraulic cements consist mainly of silicates and aluminates of lime, and can be
classified broadly as natural cements, Portland cements, and high-alumina cements.
The present chapter deals with the manufacture of Portland cement and its structure
and properties, both when unhydrated and in a hardened state. The different types of
Portland and other cements are considered in Chapter 2.
Historical note
The use of cementing materials is very old. The ancient Egyptians used calcined
impure gypsum. The Greeks and the Romans used calcined limestone and later
learned to add to lime and water, sand and crushed stone or brick and broken tiles.
This was the first concrete in history. Lime mortar does not harden under water, and
for construction under water the Romans ground together lime and a volcanic ash or
finely ground burnt clay tiles. The active silica and alumina in the ash and the tiles
combined with the lime to produce what became known as pozzolanic cement from
the name of the village of Pozzuoli, near Vesuvius, where the volcanic ash was first
found. The name ‘pozzolanic cement’ is used to this day to describe cements obtained
simply by the grinding of natural materials at normal temperature. Some of the
Roman structures in which masonry was bonded by mortar, such as the Coliseum in
Rome and the Pont du Gard near Nîmes, and concrete structures such as the Pantheon
in Rome, have survived to this day, with the cementitious material still hard and firm.
In the ruins at Pompeii, the mortar is often less weathered than the rather soft stone.
The Middle Ages brought a general decline in the quality and use of cement, and it
was only in the eighteenth century that an advance in the knowledge of cements
occurred. John Smeaton, commissioned in 1756 to rebuild the Eddystone Lighthouse,
off the Cornish coast, found that the best mortar was produced when pozzolana was
mixed with limestone containing a considerable proportion of clayey matter. By
recognizing the role of the clay, hitherto considered undesirable, Smeaton was the
first to understand the chemical properties of hydraulic lime, that is a material
obtained by burning a mixture of lime and clay.
There followed a development of other hydraulic cements, such as the ‘Roman
cement’ obtained by James Parker by calcining nodules of argillaceous limestone,
culminating in the patent for ‘Portland cement’ taken out by Joseph Aspdin, a Leeds
bricklayer, stonemason, and builder, in 1824. This cement was prepared by heating a
mixture of finely-divided clay and hard limestone in a furnace until CO2 had been
driven off; this temperature was much lower than that necessary for clinkering. The
prototype of modern cement was made in 1845 by Isaac Johnson, who burnt a mixture

7
of clay and chalk until clinkering, so that the reactions necessary for the formation of
strongly cementitious compounds took place.
The name ‘Portland cement’, given originally due to the resemblance of the colour
and quality of the hardened cement to Portland stone – a limestone quarried in Dorset
– has remained throughout the world to this day to describe a cement obtained by
intimately mixing together calcareous and argillaceous, or other silica-, alumina-, and
iron oxide-bearing materials, burning them at a clinkering temperature, and grinding
the resulting clinker. The definition of Portland cement in various standards is on
these lines, recognizing that gypsum is added after burning; nowadays, other materials
may also be added or blended (see p. 64).
Manufacture of Portland cement
From the definition of Portland cement given above, it can be seen that it is made
primarly from a calcareous material, such as limestone or chalk, and from alumina
and silica found as clay or shale. Marl, a mixture of calcareous and argillaceous
materials, is also used. Raw materials for the manufacture of Portland cement are
found in nearly all countries and cement plants operate all over the world.
The process of manufacture of cement consists essentially of grinding the raw
materials, mixing them intimately in certain proportions and burning in a large rotary
kiln at a temperature of up to about 1450 °C when the material sinters and partially
fuses into balls known as clinker. The clinker is cooled and ground to a fine powder,
with some gypsum added, and the resulting product is the commercial Portland
cement so widely used throughout the world.
Some details of the manufacture of cement will now be given, and these can be
best followed with reference to the diagrammatic representation of the process shown
in Fig. 1.1.

8
9
 Fig. 1.1. Diagrammatic representation of: (a) the wet process and (b) the dry
process of manufacture of cement
The mixing and grinding of the raw materials can be done either in water or in a
dry condition; hence the names ‘wet’ and ‘dry’ processes. The actual methods of
manufacture depend also on the hardness of the raw materials used and on their
moisture content.
Let us consider first the wet process. When chalk is used, it is finely broken up and
dispersed in water in a washmill; this is a circular pit with revolving radial arms
carrying rakes which break up the lumps of solid matter. The clay is also broken up
and mixed with water, usually in a similar washmill. The two mixtures are now
pumped so as to mix in predetermined proportions and pass through a series of
screens. The resulting cement slurry flows into storage tanks.
When limestone is used, it has to be blasted, then crushed, usually in two
progressively smaller crushers, and then fed into a ball mill with the clay dispersed in
water. There, the comminution of the limestone (to the fineness of flour) is completed,
and the resultant slurry is pumped into storage tanks. From here onwards, the process
is the same regardless of the original nature of the raw materials.
The slurry is a liquid of creamy consistency, with a water content of between 35
and 50 per cent, and only a small fraction of material – about 2 per cent – larger than a

10
90 μm (No. 170 ASTM) sieve size. There are usually a number of storage tanks in
which the slurry is kept, the sedimentation of the suspended solids being prevented by
mechanical stirrers or bubbling by compressed air. The lime content of the slurry is
governed by the proportioning of the original calcareous and argillaceous materials, as
mentioned earlier. Final adjustment in order to achieve the required chemical
composition can be made by blending slurries from different storage tanks, sometimes
using an elaborate system of blending tanks. Occasionally, for example in the world’s
northernmost plant in Norway, the raw material is a rock of such composition that it
alone is crushed and no blending is required.
Finally, the slurry with the desired lime content passes into the rotary kiln. This is
a large, refractory-lined steel cylinder, up to 8 m (or 26 ft) in diameter, sometimes as
long as 230 m (or 760 ft), slowly rotating about its axis, which is slightly inclined to
the horizontal. The slurry is fed in at the upper end while pulverized coal is blown in
by an air blast at the lower end of the kiln, where the temperature reaches about
1450 °C. The coal, which must not have too high an ash content, deserves a special
mention because typically 220 kg (500 lb) of coal is used to make one tonne of
cement. This is worth bearing in mind when considering the price of cement. Oil (of
the order of 125 litres (33 US gallons) per tonne of cement) or natural gas were also
used, but since the 1980s most oil-fired plants have been converted to coal, which is
by far the most common fuel used in most countries. It is worth noting that, because it
is burnt in the kiln, coal with a high sulfur content can be used without harmful
emissions.
The slurry, in its movement down the kiln, encounters a progressively higher
temperature. At first, the water is driven off and CO2 is liberated; further on, the dry
material undergoes a series of chemical reactions until finally, in the hottest part of
the kiln, some 20 to 30 per cent of the material becomes liquid, and lime, silica and
alumina recombine. The mass then fuses into balls, 3 to 25 mm ( to 1 in.) in diameter,
known as clinker. The clinker drops into coolers, which are of various types and often
provide means for an exchange of heat with the air subsequently used for the
combustion of the pulverized coal. The kiln has to operate continuously in order to
ensure a steady regime, and therefore uniformity of clinker, and also to reduce the
deterioration of the refractory lining. It should be noted that the flame temperature
reaches 1650 °C. The largest existing kiln in a wet-process plant produces 3600
tonnes of clinker a day. Because the manufacture of cement by the wet process is
energy intensive, new wet-process plants are no longer built.
In the dry and semi-dry processes, the raw materials are crushed and fed in the
correct proportions into a grinding mill, where they are dried and reduced in size to a
fine powder. The dry powder, called raw meal, is then pumped to a blending silo, and
final adjustment is now made in the proportions of the materials required for the
manufacture of cement. To obtain a uniform and intimate mixture, the raw meal is
blended, usually by means of compressed air inducing an upward movement of the
powder and decreasing its apparent density. The air is pumped over one quadrant of
the silo at a time, and this permits the apparently heavier material from the non-
aerated quadrants to move laterally into the aerated quadrant. Thus the aerated
material tends to behave almost like a liquid and, by aerating all quadrants in turn for
a total period of about one hour, a uniform mixture is obtained. In some cement plants,
continuous blending is used.

11
 In the semi-dry process, the blended meal is now sieved and fed into a rotating
dish called a granulator, water weighing about 12 per cent of the meal being added at
the same time. In this manner, hard pellets about 15 mm ( in.) in diameter are
formed. This is necessary, as cold powder fed direct into a kiln would not permit the
air flow and exchange of heat necessary for the chemical reactions of formation of
cement clinker.
The pellets are baked hard in a pre-heating grate by means of hot gases from the
kiln. The pellets then enter the kiln, and subsequent operations are the same as in the
wet process of manufacture. Since, however, the moisture content of the pellets is
only 12 per cent as compared with the 40 per cent moisture content of the slurry used
in the wet process, the semi-dry-process kiln is considerably smaller. The amount of
heat required is also very much lower because only some 12 per cent of moisture has
to be driven off, but additional heat has already been used in removing the original
moisture content of the raw materials (usually 6 to 10 per cent). The process is thus
quite economical, but only when the raw materials are comparatively dry. In such a
case the total coal consumption can be as little as 100 kg (220 lb) per tonne of cement.
In the dry process (see Fig. 1b), the raw meal, which has a moisture content of
about 0.2 per cent, is passed through a pre-heater, usually of a suspension type; that
means that the raw meal particles are suspended in the rising gases. Here, the raw
meal is heated to about 800 °C before being fed into the kiln. Because the raw meal
contains no moisture to be driven off and because it is already pre-heated, the kiln can
be shorter than in the wet process. The pre-heating uses the hot gas leaving the kiln.
Because that gas contains a significant proportion of the rather volatile alkalis (see
p. 9) and chlorides, a part of the gas may need to be bled off to ensure that the alkali
content of the cement is not too high.
The major part of the raw meal can be passed through a fluidized calciner (using a
separate heat source) introduced between the pre-heater and the kiln. The temperature
in the fluidized calciner is about 820 °C. This temperature is stable so that the
calcination is uniform and the efficiency of the heat exchange is high.
A part of the raw meal is fed direct into the kiln in the usual manner but, overall,
the effect of the fluidized calciner is to increase the decarbonation (dissociation of
CaCO3) of the raw meal prior to entry into the kiln and thus greatly to increase the
kiln throughput. What is probably the largest dry-process plant in the world produces
10 000 tonnes of clinker a day using a kiln 6.2 m (20 ft) in diameter and 105 m (345 ft)
long. In the U.S. more than 80% of cement production uses the dry process.
It should be stressed that all processes require an intimate mixture of the raw
materials because a part of the reactions in the kiln must take place by diffusion in
solid materials, and a uniform distribution of materials is essential to ensure a uniform
product.
On exit from the kiln, regardless of the type of process, the clinker is cooled, the
heat being used to preheat the combustion air. The cool clinker, which is
characteristically black, glistening, and hard, is interground with gypsum in order to
prevent flash setting of the cement. The grinding is done in a ball mill consisting of
several compartments with progressively smaller steel balls, sometimes preceded by
passing through a roll press. In most plants, a closed-circuit grinding system is used:
the cement discharged by the mill is passed through a separator, fine particles being
removed to the storage silo by an air current, while the coarser particles are passed
through the mill once again. Closed-circuit grinding avoids the production of a large

12
amount of excessively fine material or of a small amount of too coarse material, faults
often encountered with open-circuit grinding. Small quantities of grinding aids such
as ethylene glycol or propylene glycol are used. Information about grinding aids is
given by Massazza and Testolin.1.90 The performance of a ball mill can be improved by
pre-grinding the clinker in a horizontal impact crusher.
Once the cement has been satisfactorily ground, when it will have as many as 1.1 ×
10 particles per kg (5 × 1011 per lb), it is ready for transport in bulk. Less commonly,
12

the cement is packed in bags or drums. However, some types of cement, such as white,
hydrophobic, expansive, regulated-set, oil-well, and high-alumina, are always packed
in bags or drums. A standard bag in the United Kingdom contains 50 kg (110 lb) of
cement; a US sack weighs 94 lb (42.6 kg); other bag sizes are also used. Bags of 25
kg are becoming popular.
Except when the raw materials necessitate the use of the wet process, the dry
process is used nowadays in order to minimize the energy required for burning.
Typically, the burning process represents 40 to 60 per cent of the production cost,
while the extraction of raw materials for the manufacture of cement represents only
10 per cent of the total cost of cement.
Around 1990, the average energy consumption in the United States for the
production of 1 tonne of cement by the dry process was 1.6 MWh. In modern plants,
this figure is much reduced, being below 0.8 MWh in Austria.1.96 Electricity
consumption, which accounts for some 6 to 8 per cent of total energy used, is
typically of the order: 10 kWh for crushing the raw materials, 28 kWh in the raw meal
preparation, 24 kWh in burning, and 41 kWh in grinding.1.18 The capital cost of
installation of a cement plant is very high: nearly US$200 per tonne of cement
produced per annum.
In addition to the main processes, there are also other processes of manufacture of
cement, of which one, using gypsum instead of lime, perhaps deserves mention.
Gypsum, clay and coke with sand and iron oxide are burnt in a rotary kiln, the end
products being Portland cement and sulfur dioxide which is further converted into
sulfuric acid.
In areas where only a small cement production is required or where investment
capital is limited, a vertical kiln of the Gottlieb type can be used. This fires nodules of
raw meal and fine coal powder combined, and produces agglomerated clinker which
is then broken up. A single kiln, 10 m (33 ft) high, produces up to 300 tonnes of
cement a day. China used several thousand of such kilns, but now has a very large
modern cement industry, producing 1000 million tonnes per annum.
Chemical composition of Portland cement
We have seen that the raw materials used in the manufacture of Portland cement
consist mainly of lime, silica, alumina and iron oxide. These compounds interact with
one another in the kiln to form a series of more complex products and, apart from a
small residue of uncombined lime which has not had sufficient time to react, a state of
chemical equilibrium is reached. However, equilibrium is not maintained during
cooling, and the rate of cooling will affect the degree of crystallization and the
amount of amorphous material present in the cooled clinker. The properties of this
amorphous material, known as glass, differ considerably from those of crystalline
compounds of a nominally similar chemical composition. Another complication arises

13
from the interaction of the liquid part of the clinker with the crystalline compounds
already present.
Nevertheless, cement can be considered as being in frozen equilibrium, i.e. the
cooled products are assumed to reproduce the equilibrium existing at the clinkering
temperature. This assumption is, in fact, made in the calculation of the compound
composition of commercial cements: the ‘potential’ composition is calculated from
the measured quantities of oxides present in the clinker as if full crystallization of
equilibrium products had taken place.
Four compounds are usually regarded as the major constituents of cement: they are
listed in Table 1.1, together with their abbreviated symbols. This shortened notation,
used by cement chemists, describes each oxide by one letter, viz.: CaO = C; SiO2 = S;
Al2O3 = A; and Fe2O3 = F. Likewise, H2O in hydrated cement is denoted by H, and
SO3 by .

Table 1.1. Main Compounds of Portland Cement

In reality, the silicates in cement are not pure compounds, but contain minor oxides
in solid solution. These oxides have significant effects on the atomic arrangements,
crystal form and hydraulic properties of the silicates.
The calculation of the potential composition of Portland cement is based on the
work of R. H. Bogue and others, and is often referred to as ‘Bogue composition’.
Bogue’s1.2 equations for the percentages of main compounds in cement are given
below. The terms in brackets represent the percentage of the given oxide in the total
mass of cement.
C3S = 4.07(CaO) – 7.60(SiO2) – 6.72(Al2O3) – 1.43(Fe2O3) – 2.85(SO3)
C2S = 2.87(SiO2) – 0.75(3CaO.SiO2)
C3A = 2.65(Al2O3) – 1.69(Fe2O3)
C4AF = 3.04(Fe2O3).
There are also other methods of calculating the composition,1.1 but the subject is not
considered to be within the scope of this book. We should note, however, that the
Bogue composition underestimates the C3S content (and overestimates C2S) because
other oxides replace some of the CaO in C3S; as already stated, chemically pure C3S
and C2S do not occur in Portland cement clinker.
A modification of the Bogue composition which takes into account the presence of
substituent ions in the nominally pure main compounds has been developed by
Taylor1.84 for the rapidly cooled clinkers produced in modern cement plants.
In addition to the main compounds listed in Table 1.1, there exist minor
compounds, such as MgO, TiO2, Mn2O3, K2O and Na2O; they usually amount to not
more than a few per cent of the mass of cement. Two of the minor compounds are of

14
particular interest: the oxides of sodium and potassium, Na2O and K2O, known as the
alkalis (although other alkalis also exist in cement). They have been found to react
with some aggregates, the products of the reaction causing disintegration of the
concrete, and have also been observed to affect the rate of the gain of strength of
cement.1.3 It should, therefore, be pointed out that the term ‘minor compounds’ refers
primarily to their quantity and not necessarily to their importance. The quantity of the
alkalis and of Mn2O3 can be rapidly determined using a spectrophotometer.
The compound composition of cement has been established largely on the basis of
studies of phase equilibria of the ternary systems C–A–S and C–A–F, and the
quaternary system C–C2S–C5A3–C4AF, and others. The course of melting or
crystallization was traced and the compositions of liquid and solid phases at any
temperature were computed. In addition to the methods of chemical analysis, the
actual composition of clinker can be determined by a microscope examination of
powder preparations and their identification by the measurement of the refractive
index. Polished and etched sections can be used both in reflected and transmitted light.
Other methods include the use of X-ray powder diffraction to identify the crystalline
phases and also to study the crystal structure of some of the phases, and of differential
thermal analysis; quantitative analysis is also possible, but complicated calibrations
are involved.1.68 Modern techniques include phase analysis through a scanning electron
microscope and image analysis through an optical microscope or a scanning electron
microscope.
Estimating the composition of cement is aided by more rapid methods of
determining the elemental composition, such as X-ray fluorescence, X-ray
spectrometry, atomic absorption, flame photometry, and electron probe micro-
analysis. X-ray diffractometry is useful in the determination of free lime, i.e. CaO as
distinct from Ca(OH)2, and this is convenient in controlling the kiln performance.1.67
C3S, which is normally present in the largest amounts, occurs as small,
equidimensional colourless grains. On cooling below 1250 °C, it decomposes slowly
but, if cooling is not too slow, C3S remains unchanged and is relatively stable at
ordinary temperatures.
C2S is known to have three, or possibly even four, forms. α-C2S, which exists at
high temperatures, inverts to the β-form at about 1450 °C. β-C2S undergoes further
inversion to γ-C2S at about 670 °C but, at the rates of cooling of commercial cements,
β-C2S is preserved in the clinker. β-C2S forms rounded grains, usually showing
twinning.
C3A forms rectangular crystals, but C3A in frozen glass forms an amorphous
interstitial phase.
C4AF is really a solid solution ranging from C2F to C6A2F, but the description C4AF
is a convenient simplification.1.4
The actual proportions of the various compounds vary considerably from cement
to cement, and indeed different types of cement are obtained by suitable proportioning
of the raw materials. In the United States, an attempt was at one time made to control
the properties of cements required for different purposes by specifying the limits of
the four major compounds, as calculated from the oxide analysis. This procedure
would cut out numerous physical tests normally performed, but unfortunately the
calculated compound composition is not sufficiently accurate, nor does it take into
account all the relevant properties of cement, and cannot therefore serve as a
substitute for direct testing of the required properties.

15
 A general idea of the composition of cement can be obtained from Table 1.2,
which gives the oxide composition limits of Portland cements. Table 1.3 gives the
oxide composition of a typical cement of the 1960s and the calculated compound
composition,1.5 obtained by means of Bogue’s equations on p. 9.

Table 1.2. Usual Composition Limits of Portland Cement

Table 1.3. Oxide and Compound Compositions of a Typical Portland Cement of

the 1960s1.5

Two terms used in Table 1.3 require explanation. The insoluble residue,
determined by treating with hydrochloric acid, is a measure of adulteration of cement,
largely arising from impurities in gypsum. British Standard BS 12 : 1996 (withdrawn)
limits the insoluble residue to 1.5 per cent of the mass of cement. European Standard
BS EN 197-1 : 2000, which allows a 5 per cent content of a filler (see p. 88), limits
the insoluble residue to 5 per cent of the mass of the cement exclusive of the filler.
The loss on ignition shows the extent of carbonation and hydration of free lime and
free magnesia due to the exposure of cement to the atmosphere. The maximum loss
on ignition (at 1000 °C) permitted by BS EN 197-1 : 2000 is 5 per cent and by ASTM
C 150-09 is 3 per cent except for Type I cement (2.5 per cent); 4 per cent is
acceptable for cements in the tropics. Because hydrated free lime is innocuous (see
p. 51), for a given free lime content of cement, a greater loss on ignition is really

16
advantageous. With cements containing a calcareous filler, a higher limit on the loss
on ignition is necessary: 5 per cent of the mass of the cement nucleus is allowed by
BS EN 197-1 : 2000.
It is interesting to observe the large influence of a change in the oxide composition
on the compound composition of cement. Some data of Czernin’s1.5 are given in Table
1.4; column (1) shows the composition of a fairly typical rapid-hardening cement. If
the lime content is decreased by 3 per cent, with corresponding increases in the other
oxides (column (2)), a considerable change in the C3S : C2S ratio results. Column (3)
shows a change of per cent in the alumina and iron contents compared with the
cement of column (1). The lime and silica contents are unaltered and yet the ratio of
the two silicates, as well as the contents of C3A and C4AF, is greatly affected. It is
apparent that the significance of the control of the oxide composition of cement
cannot be over-emphasized. Within the usual range of ordinary and rapid-hardening
Portland cements the sum of the contents of the two silicates varies only within
narrow limits, so that the variation in composition depends largely on the ratio of CaO
to SiO2 in the raw materials.

Table 1.4. Influence of Change in Oxide Composition on the Compound

Composition1.5

In some countries in the European Union, there is a limit on soluble hexavalent

chromium usually of 2 ppm of mass dry cement. Contact with excess chromium in
fresh concrete may lead to dermatitis.
It may be convenient at this stage to summarize the pattern of formation and
hydration of cement; this is shown schematically in Fig. 1.2.

17
 Fig. 1.2. Schematic representation of the formation and hydration of Portland
cement
Hydration of cement
The reactions by virtue of which Portland cement becomes a bonding agent take place
in a water–cement paste. In other words, in the presence of water, the silicates and
aluminates listed in Table 1.1 form products of hydration which in time produce a
firm and hard mass – the hydrated cement paste.
There are two ways in which compounds of the type present in cement can react
with water. In the first, a direct addition of some molecules of water takes place, this
being a true reaction of hydration. The second type of reaction with water is
hydrolysis. It is convenient and usual, however, to apply the term hydration to all
reactions of cement with water, i.e. to both true hydration and hydrolysis.
Le Chatelier was the first to observe, about 130 years ago, that the products of
hydration of cement are chemically the same as the products of hydration of the
individual compounds under similar conditions. This was later confirmed by
Steinour1.6 and by Bogue and Lerch,1.7 with the proviso that the products of reaction
may influence one another or may themselves interact with the other compounds in
the system. The two calcium silicates are the main cementitious compounds in cement,
and the physical behaviour of cement during hydration is similar to that of these two
compounds alone.1.8 The hydration of the individual compounds will be described in
more detail in the succeeding sections.
The products of hydration of cement have a very low solubility in water as shown
by the stability of the hydrated cement paste in contact with water. The hydrated
cement bonds firmly to the unreacted cement, but the exact way in which this is

18
achieved is not certain. It is possible that the newly produced hydrate forms an
envelope which grows from within by the action of water that has penetrated the
surrounding film of hydrate. Alternatively, the dissolved silicates may pass through
the envelope and precipitate as an outer layer. A third possibility is for the colloidal
solution to be precipitated throughout the mass after the condition of saturation has
been reached, the further hydration continuing within this structure.
Whatever the mode of precipitation of the products of hydration, the rate of
hydration decreases continuously, so that even after a long time there remains an
appreciable amount of unhydrated cement. For instance, after 28 days in contact with
water, grains of cement have been found to have hydrated to a depth of only
4 μm,1.9 and 8 μm after a year. Powers1.10 calculated that complete hydration under
normal conditions is possible only for cement particles smaller than 50 μm, but full
hydration has been obtained by grinding cement in water continuously for five days.
Microscopic examination of hydrated cement shows no evidence of channelling of
water into the grains of cement to hydrate selectively the more reactive compounds
(e.g. C3S) which may lie in the centre of the particle. It would seem then, that
hydration proceeds by a gradual reduction in the size of the cement particle. In fact,
unhydrated grains of coarse cement were found to contain C3S as well as C2S at the
age of several months,1.11 and it is probable that small grains of C2S hydrate before the
hydration of large grains of C3S has been completed. The various compounds in
cement are generally intermixed in all grains, and some investigations have suggested
that the residue of a grain after a given period of hydration has the same percentage
composition as the whole of the original grain.1.12 However, the composition of the
residue does change throughout the period of cement hydration,1.49 and especially
during the first 24 hours selective hydration may take place.
The main hydrates can be broadly classified as calcium silicate hydrates and
tricalcium aluminate hydrate. C4AF is believed to hydrate into tricalcium aluminate
hydrate and an amorphous phase, probably CaO.Fe2O3.aq. It is possible also that some
Fe2O3 is present in solid solution in the tricalcium aluminate hydrate.
The progress of hydration of cement can be determined by different means, such as
the measurement of: (a) the amount of Ca(OH)2 in the paste; (b) the heat evolved by
hydration; (c) the specific gravity of the paste; (d) the amount of chemically combined
water; (e) the amount of unhydrated cement present (using X-ray quantitative
analysis); and (f) also indirectly from the strength of the hydrated paste.
Thermogravimetric techniques and continuous X-ray diffraction scanning of wet
pastes undergoing hydration1.50 can be used in studying early reactions. The
microstructure of hydrated cement paste can also be studied by back-scattered
electron imaging in a scanning electron microscope.
Calcium silicate hydrates
The rates of hydration of C3S and C2S in a pure state differ considerably, as shown
in Fig. 1.3. When the various compounds are present all together in cement, their rates
of hydration are affected by compound interactions. In commercial cements, the
calcium silicates contain small impurities of some of the oxides present in the clinker.
The ‘impure’ C3S is known as alite and the ‘impure’ C2S as belite. These impurities
have a strong effect on the properties of the calcium silicate hydrates (see p. 48).

19
 Fig. 1.3. Typical development of hydration of pure compounds1.47
When hydration takes place in a limited amount of water, as is the case in cement
paste, in mortar or in concrete, C3S is believed to undergo hydrolysis producing a
calcium silicate of lower basicity, ultimately C3S2H3, with the released lime separating
out as Ca(OH)2. There exists, however, some uncertainty as to whether C3S and C2S
result ultimately in the same hydrate. It would appear to be so from considerations of
the heat of hydration1.6 and of the surface area of the products of hydration,1.13 but
physical observations indicate that there may be more than one – possibly several –
distinct calcium silicate hydrates. The C : S ratio would be affected if some of the
lime were absorbed or held in solid solution, and there is strong evidence that the
ultimate product of hydration of C2S has a lime/silica ratio of 1.65. This may be due to
the fact that the hydration of C3S is controlled by the rate of diffusion of ions through
the overlying hydrate films while the hydration of C2S is controlled by its slow
intrinsic rate of reaction.1.14 Furthermore, temperature may affect the products of
hydration of the two silicates because the permeability of the gel is affected by
temperature.
The C : S ratio has not been unequivocally determined because different test
methods yield different results.1.74 The variation can be as wide as 1.5 by chemical
extraction and 2.0 by thermogravimetric method.1.66 Electron-optical measurements
also yield low values of the C : S ratio.1.72 The ratio also varies with time and is
influenced by the presence of other elements or compounds in the cement. Nowadays,
the calcium silicate hydrates are broadly described as C–S–H, and the C : S ratio is
believed to be probably near 2.1.19 Because the crystals formed by hydration are
imperfect and extremely small, the mole ratio of water to silica need not be a whole
number. C–S–H usually contains small amounts of Al, Fe, Mg, and other ions. At one
time, C–S–H was referred to as tobermorite gel because of a structural similarity to a
mineral of this name, but this may not be correct,1.60 and this description is now rarely
used.
Making the approximate assumption that C3S2H3 is the final product of hydration of
both C3S and C2S, the reactions of hydration can be written (as a guide, although not
as exact stoichiometric equations) as follows.

20
 For C3S:

2C3S + 6H → C3S2H3 + 3Ca(OH)2.

The corresponding masses involved are:

100 + 24 → 75 + 49.

For C2S:

2C2S + 4H → C3S2H3 + Ca(OH)2.

The corresponding masses are:

100 + 21 → 99 + 22.

Thus, on a mass basis, both silicates require approximately the same amount of
water for their hydration, but C3S produces more than twice as much Ca(OH)2 as is
formed by the hydration of C2S.
The physical properties of the calcium silicate hydrates are of interest in
connection with the setting and hardening properties of cement. These hydrates
appear amorphous but electron microscopy shows their crystalline character. It is
interesting to note that one of the hydrates believed to exist, denoted by Taylor1.15 as
CSH(I), has a layer structure similar to that of some clay minerals, e.g.
montmorillonite and halloysite. The individual layers in the plane of the a and b axes
are well crystallized while the distances between them are less rigidly defined. Such a
lattice would be able to accommodate varying amounts of lime without fundamental
change – a point relevant to the varying lime/silica ratios mentioned above. In fact,
powder diagrams have shown that lime in excess of one molecule per molecule of
silica is held in a random manner.1.15 Steinour1.16 described this as a merger of solid
solution and adsorption.
Calcium silicates do not hydrate in the solid state but the anhydrous silicates
probably first pass into solution and then react to form less soluble hydrated silicates
which separate out of the supersaturated solution.1.17 This is the type of mechanism of
hydration first envisaged by Le Chatelier in 1881.
Studies by Diamond1.60 indicate that the calcium silicate hydrates exist in a variety
of forms: fibrous particles, flattened particles, a reticular network, irregular grains, all
rather difficult to define. However, the predominant form is that of fibrous particles,
possibly solid, possibly hollow, sometimes flattened, sometimes branching at the ends.
Typically, they are 0.5 μm to 2 μm long and less than 0.2 μm across. This is not a
precise picture, but the structure of calcium silicate hydrates is too disordered to be
established by the existing techniques, including a combination of the scanning
electron microscope and energy dispersive X-ray spectrometer.
The hydration of C3S to a large extent characterizes the behaviour of cement and a
description of the latter may be appropriate. Hydration does not proceed at a steady
rate or even at a steadily changing rate. The initial rapid release of calcium hydroxide
into the solution leaves an outer layer of calcium silicate hydrate, perhaps 10 nm
thick.1.61 This layer impedes further hydration so that, for some time thereafter, very
little hydration takes place.
As the hydration of cement is an exothermic reaction, the rate of evolution of heat
is an indication of the rate of hydration. This shows that there are three peaks in the

21
rate of hydration in the first three days or so, from the time when the dry cement first
comes into contact with water. Figure 1.4 shows a plot of the rate of evolution of heat
against time.1.81 We can see the first peak, which is very high, and which corresponds
to the initial hydration at the surface of the cement particles, largely involving C3A.
The duration of this high rate of hydration is very short, and there follows a so-
called dormant period, sometimes called also an induction period, during which the
rate is very low. This period lasts one or two hours during which the cement paste is
workable.

Fig. 1.4. Rate of evolution of heat of Portland cement with a water/cement ratio
of 0.4.181 The first peak of 3200 J/s kg is off the diagram
Eventually, the surface layer is broken down, possibly by an osmotic mechanism
or by the growth of the crystals of calcium hydroxide. The rate of hydration (and
therefore of heat evolution) increases fairly slowly and the products of hydration of
individual grains come into contact with one another; setting then occurs. The rate of
heat evolution reaches a second peak, typically at the age of about 10 hours, but
sometimes as early as 4 hours.
Following this peak, the rate of hydration slows down over a long period, the
diffusion through the pores in the products of hydration becoming the controlling
factor.1.62 With most, but not all, cements, there is a renewed increase in the rate of
hydration up to a third, lower, peak at the age of between 18 and 30 hours. This peak
is related to a renewed reaction of C3A, following the exhaustion of gypsum.
The advent of the second peak is accelerated by the presence of the alkalis, by a
higher fineness of the cement particles, and by an increase in temperature.
Because of the similarity in the progress of hydration of neat calcium silicates and
of commercial Portland cements, they show similar strength development.1.20 A
considerable strength is possessed long before the reactions of hydration are complete
and it would thus seem that a small amount of the hydrate binds together the
unhydrated remainder; further hydration results in little increase in strength.
Ca(OH)2 liberated by the hydrolysis of the calcium silicates forms thin hexagonal
plates, often tens of micrometres across, but later they merge into a massive deposit.1.60
Tricalcium aluminate hydrate and the action of gypsum

22
The amount of C3A present in most cements is comparatively small but its behaviour
and structural relationship with the other phases in cement make it of interest. The
tricalcium aluminate hydrate forms a prismatic dark interstitial material, possibly with
other substances in solid solution, and is often in the form of flat plates individually
surrounded by the calcium silicate hydrates.
The reaction of pure C3A with water is very violent and leads to immediate
stiffening of the paste, known as flash set. To prevent this from happening, gypsum
(CaSO4.2H2O) is added to cement clinker. Gypsum and C3A react to form insoluble
calcium sulfoaluminate (3CaO.Al2O3.3CaSO4.32H2O), but eventually tricalcium
aluminate hydrate is formed, although this is preceded by a metastable
3CaO.Al2O3.CaSO4.12H2O, produced at the expense of the original high-sulfate
calcium sulfoaluminate.1.6 As more C3A comes into solution, the composition changes,
the sulfate content decreasing continuously. The rate of reaction of the aluminate is
high and, if this readjustment in composition is not rapid enough, direct hydration of
C3A is likely. In particular, the first peak in the rate of heat development, normally
observed within five minutes of adding water to cement, means that some calcium
aluminate hydrate is formed directly during that period, the conditions for the
retardation by gypsum not yet having been established.
Instead of gypsum, other forms of calcium sulfate can be used in the manufacture
of cement: hemihydrate (CaSO4. H2O) or anhydrite (CaSO4).
There is some evidence that the hydration of C3A is retarded by Ca(OH)2 liberated
by the hydrolysis of C3S.1.62 This occurs due to the fact that Ca(OH)2 reacts with C3A
and water to form C4AH19, which forms a protective coating on the surface of
unhydrated grains of C3A. It is also possible that Ca(OH)2 decreases the concentration
of aluminate ions in the solution, thus slowing down the rate of hydration of C3A.1.62
The stable form of the calcium aluminate hydrate ultimately existing in the
hydrated cement paste is probably the cubic crystal C3AH6, but it is possible that
hexagonal C4AH12 crystallizes out first and later changes to the cubic form. Thus the
final form of the reaction can be written:

C3A + 6H → C3AH6.

This again is an approximation and not a stoichiometric equation.

The molecular weights show that 100 parts of C3A react with 40 parts of water by
mass, which is a much higher proportion of water than that required by the silicates.
The presence of C3A in cement is undesirable: it contributes little or nothing to the
strength of cement except at early ages and, when hardened cement paste is attacked
by sulfates, expansion due to the formation of calcium sulfoaluminate from C3A may
result in a disruption of the hardened paste. However, C3A acts as a flux and thus
reduces the temperature of burning of clinker and facilitates the combination of lime
and silica; for these reasons, C3A is useful in the manufacture of cement. C4AF also
acts as a flux. It may be noted that if some liquid were not formed during burning, the
reactions in the kiln would progress much more slowly and would probably be
incomplete. On the other hand, a higher C3A content increases the energy required to
grind the clinker.
A positive effect of C3A is its binding capacity of chlorides (see p. 571).

23
 Gypsum reacts not only with C3A: with C4AF it forms calcium sulfoferrite as well
as calcium sulfoaluminate, and its presence may accelerate the hydration of the
silicates.
The amount of gypsum added to the cement clinker has to be very carefully
watched; in particular, an excess of gypsum leads to an expansion and consequent
disruption of the set cement paste. The optimum gypsum content is determined by
observation of the generation of the heat of hydration. As already mentioned, the first
peak in the rate of heat evolution is followed by a second peak some 4 to 10 hours
after the water had been added to cement, and with the correct amount of gypsum
there should be little C3A available for reaction after all the gypsum has combined,
and no further peak in the heat liberation should occur. Thus, an optimum gypsum
content leads to a desirable rate of early reaction and prevents local high
concentration of products of hydration (see p. 362). In consequence, the size of pores
in hydrated cement paste is reduced and strength is increased.1.78
The amount of gypsum required increases with the C3A content and also with the
alkali content of the cement. Increasing the fineness of cement has the effect of
increasing the quantity of C3A available at early stages, and this raises the gypsum
requirement. A test for the optimum SO3 content in Portland cement was prescribed
by ASTM C 543-84 (discontinued). The optimization is based on a 1-day strength,
which usually also produces the lowest shrinkage.
The amount of gypsum added to cement clinker is expressed as the mass of
SO3 present; this is limited by European Standard BS EN 197-1 : 2000 to a maximum
of 3.5 per cent, but in some cases higher percentages are permitted. The chemically
relevant SO3 is the soluble sulfate contributed by gypsum and not that from high-
sulfur fuel, which is bound in the clinker; this is why the current total SO3 limit is
higher than in the past. The maximum values of SO3 laid down in ASTM C 150-09
depend on the content of C3A, and are higher in rapid-hardening cement.
Setting
This is the term used to describe the stiffening of the cement paste, although the
definition of the stiffness of the paste which is considered set is somewhat arbitrary.
Broadly speaking, setting refers to a change from a fluid to a rigid stage. Although,
during setting, the paste acquires some strength, for practical purposes it is important
to distinguish setting from hardening, which refers to the gain of strength of a set
cement paste.
In practice, the terms initial set and final set are used to describe arbitrarily chosen
stages of setting. The method of measurement of these setting times is described on
p. 49.
It seems that setting is caused by a selective hydration of cement compounds: the
two first to react are C3A and C3S. The flash-setting properties of the former were
mentioned in the preceding section but the addition of gypsum delays the formation of
calcium aluminate hydrate, and it is thus C3S that sets first. Pure C3S mixed with water
also exhibits an initial set but C2S stiffens in a more gradual manner.
In a properly retarded cement, the framework of the hydrated cement paste is
established by the calcium silicate hydrate, while, if C3A were allowed to set first, a
rather porous calcium aluminate hydrate would form. The remaining cement
compounds would then hydrate within this porous framework and the strength
characteristics of the cement paste would be adversely affected.

24
 Apart from the rapidity of formation of crystalline products, the development of
films around cement grains and a mutual coagulation of components of the paste have
also been suggested as factors in the development of set.
At the time of the final set, there is a sharp drop in the electrical conductivity of the
cement paste, and attempts have been made to measure setting by electrical means.
The setting time of cement decreases with a rise in temperature, but above about
30 °C (85 °F) a reverse effect may be observed.1.1 At low temperatures setting is
retarded.
False set
False set is the name given to the abnormal premature stiffening of cement within a
few minutes of mixing with water. It differs from flash set in that no appreciable heat
is evolved, and remixing the cement paste without addition of water restores plasticity
of the paste until it sets in the normal manner and without a loss of strength.
Some of the causes of false set are to be found in the dehydration of gypsum when
interground with too hot a clinker: hemihydrate (CaSO4. H2O) or anhydrite (CaSO4)
are formed and when the cement is mixed with water these hydrate to form needle-
shaped crystals of gypsum. Thus what is called ‘plaster set’ takes place with a
resulting stiffening of the paste.
Another cause of false set may be associated with the alkalis in the cement. During
storage they may carbonate, and alkali carbonates react with Ca(OH)2, liberated by the
hydrolysis of C3S, to form CaCO3. This precipitates and induces a rigidity of the paste.
It has also been suggested that false set can be due to the activation of C3S by
aeration at moderately high humidities. Water is adsorbed on the grains of cement,
and these freshly activated surfaces can combine very rapidly with more water during
mixing: this rapid hydration would produce false set.1.21
Laboratory tests at cement plants generally ensure that cement is free from false set.
If, however, false set is encountered, it can be dealt with by remixing the concrete
without adding any water. Although this is not easy, workability will be improved and
the concrete can be placed in the normal manner.
Fineness of cement
It may be recalled that one of the last steps in the manufacture of cement is the
grinding of clinker mixed with gypsum. Because hydration starts at the surface of the
cement particles, it is the total surface area of cement that represents the material
available for hydration. Thus, the rate of hydration depends on the fineness of the
cement particles and, for a rapid development of strength, high fineness is necessary
(see Fig. 1.5); the long-term strength is not affected. A higher early rate of hydration
means, of course, also a higher rate of early heat evolution.

25
 Fig. 1.5. Relation between strength of concrete at different ages and fineness of
cement1.43
On the other hand, the cost of grinding to a higher fineness is considerable, and
also the finer the cement the more rapidly it deteriorates on exposure to the
atmosphere. Finer cement leads to a stronger reaction with alkali-reactive
aggregate,1.44 and makes the cement paste, though not necessarily concrete, exhibit a
higher shrinkage and a greater proneness to cracking. However, fine cement bleeds
less than a coarser one.
An increase in fineness increases the amount of gypsum required for proper
retardation because, in a finer cement, more C3A is available for early hydration. The
water content of a paste of standard consistency is greater the finer the cement, but
conversely an increase in fineness of cement slightly improves the workability of a
concrete mix. This anomaly may be due partly to the fact that the tests for consistency
of cement paste and workability measure different properties of fresh paste; also,
accidental air affects the workability of cement paste, and cements of different
fineness may contain different amounts of air.
We can see then that fineness is a vital property of cement and has to be carefully
controlled. The fraction of cement retained on a 45 μm (No. 325 ASTM) test sieve
can be determined using ASTM C 430-08. (For size of openings of different sieves
see Table 3.14.) This would ensure that the cement does not contain an excess of large
grains which, because of their comparatively small surface area per unit mass, would
play only a small role in the process of hydration and development of strength.
However, the sieve test gives no information on the size of grains smaller than
45 μm (No. 325 ASTM) sieve, and it is the finer particles that play the greatest part in
the early hydration.

26
 For this reason, modern standards prescribe a test for fineness by determination of
the specific surface of cement expressed as the total surface area in square metres per
kilogram. A direct approach is to measure the particle size distribution by
sedimentation or elutriation: these methods are based on the dependence of the rate of
free fall of particles on their diameter. Stokes’ law gives the terminal velocity of fall
under gravity of a spherical particle in a fluid medium; the cement particles are, in
fact, not spherical. This medium must of course be chemically inert with respect to
cement. It is also important to achieve a satisfactory dispersion of cement particles as
partial flocculation would produce a decrease in the apparent specific surface.
A development of these methods is the Wagner turbidimeter used in the United
States (ASTM C 115-10). In this test, the concentration of particles in suspension at a
given level in kerosene is determined using a beam of light, the percentage of light
transmitted being measured by a photocell. The turbidimeter gives generally
consistent results, but an error is introduced by assuming a uniform size distribution
of particles smaller than 7.5 μm. It is precisely these finest particles that contribute
most to the specific surface of cement and the error is especially significant with the
finer cements used nowadays. However, an improvement on the standard method is
possible if the concentration of particles 5 μm in size is determined and a modification
of calculations is made.1.51 A typical curve of particle size distribution is shown in Fig.
1.6, which gives also the corresponding contribution of these particles to the total
surface area of the sample. As mentioned on p. 7, the particle size distribution
depends on the method of grinding and varies, therefore, from plant to plant.

Fig. 1.6. Example of particle size distribution and cumulative surface area
contributed by particles up to any given size for 1 g of cement

27
 It must be admitted, however, that it is not quite clear what is a ‘good’ grading of
cement: should all the particles be of the same size or should their distribution be such
that they are able to pack densely? It is now believed that, for a given specific surface
of cement, early strength development is better if at least 50 per cent of the particles
lie between 3 and 30 μm, with correspondingly fewer very fine and fewer very coarse
particles. An even higher proportion of particles in the range of 3 to 30 μm, up to 95
per cent, is believed to lead to an improved early strength and also to a good ultimate
strength of concrete made with such a cement. To achieve such a controlled particle
size distribution it is necessary to use high-efficiency classifiers in closed-circuit
grinding of clinker. These classifiers reduce the amount of energy used in grinding.1.80
The reason for the beneficial effect of middle-size particles may be found in the
test results of Aïtcin et al.1.91 who found that grinding of cement results in a certain
amount of compound segregation. Specifically, particles smaller than 4 μm are very
rich in SO3 and rich in the alkalis; particles coarser than 30 μm contain a large
proportion of C2S, while the particles between 4 and 30 μm are rich in C3S.
It should be noted, however, that there is no simple relation between strength and
cement particle size distribution: for example, weathered, partly hydrated clinker,
after grinding, results in cement with a misleadingly high apparent surface area.
The specific surface of cement can also be determined by the air permeability
method, using an apparatus developed by Lea and Nurse. The method is based on the
relation between the flow of a fluid through a granular bed and the surface area of the
particles comprising the bed. From this, the surface area per unit mass of the bed
material can be related to the permeability of a bed of a given porosity, i.e. containing
a fixed volume of pores in the total volume of the bed.
The permeability apparatus is shown diagrammatically in Fig. 1.7. Knowing the
density of cement, the mass required to make a bed of porosity of 0.475 and 10 mm
thick can be calculated. This amount of cement is placed in a cylindrical container, a
stream of dry air is passed through the cement bed at a constant velocity, and the
resulting pressure drop is measured by a manometer connected to the top and bottom
of the bed. The rate of airflow is measured by a flowmeter consisting of a capillary
placed in the circuit and a manometer across its ends.

28
 Fig. 1.7. Lea and Nurse permeability apparatus
An equation developed by Carman gives the specific surface in square centimetres
per gram as

where ρ = density of cement (g/cm3)

ε = porosity of cement bed (0.475 in the BS test)
A = cross-sectional area of the bed (5.066 cm2)
L = height of the bed (1 cm)
h1 = pressure drop across the bed
h2 = pressure drop across the flowmeter capillary (between 25 and 55 cm of
kerosene)
and K = the flowmeter constant.
For a given aparatus and porosity the expression simplifies to

where K1 is a constant.

29
 In the United States and nowadays in Europe, a modification of the Lea and Nurse
method, developed by Blaine, is used; the method is prescribed by ASTM C 204-07
and by BS EN 196-6 : 2010. Here, the air does not pass through the bed at a constant
rate but a known volume of air passes at a prescribed average pressure, the rate of
flow diminishing steadily. The time t for the flow to take place is measured, and for a
given apparatus and a standard porosity of 0.500, the specific surface is given by

where K2 is a constant.
The Lea and Nurse and the Blaine methods give values of specific surface in close
agreement with one another but very much higher than the value obtained by the
Wagner method. This is due to Wagner’s assumptions about the size distribution of
particles below 7.5 μm, mentioned earlier. The actual distribution in this range is such
that the average value of 3.75 μm, assumed by Wagner, underestimates the surface
area of these particles. In the air permeability method, the surface area of all particles
is measured directly, and the resulting value of the specific surface is about 1.8 times
higher than the value calculated by the Wagner method. The actual range of the
conversion factor varies between 1.6 and 2.2, depending on the fineness of the cement
and its gypsum content.
Either method gives a good picture of the relative variation in the fineness of
cement, and for practical purposes this is sufficient. The Wagner method is somewhat
more informative in that it gives an indication of the particle size distribution. An
absolute measurement of the specific surface can be obtained by the nitrogen
adsorption method, based on the work of Brunauer, Emmett and Teller.1.45 While in the
air-permeability methods only the continuous paths through the bed of cement
contribute to the measured area, in the nitrogen adsorption method the ‘internal’ area
is also accessible to the nitrogen molecules. For this reason, the measured value of the
specific surface is considerably higher than that determined by the air permeability
methods. Some typical values are given in Table 1.5.

Table 1.5. Specific Surface of Cement Measured by Different Methods1.1

The surface area of powders much finer than Portland cement, such as silica fume
and fly ash, cannot be determined by the air permeability method but requires the use
of gas adsorption methods, such as the nitrogen adsorption method. The latter is time
consuming, and it may be preferable to use mercury intrusion porosimetry; 1.69 however,
this technique has not yet been generally accepted.
Modern specifications no longer lay down minimum values of the specific surface
of Portland cement, this being indirectly controlled by the early strength requirement
where appropriate. It may be useful to state, however, that a typical ordinary Portland
cement would have a specific surface of about 350 or 380 m2/kg; the specific surface
of rapid-hardening Portland cement is typically higher.

30
Structure of hydrated cement
Many of the mechanical properties of hardened cement and concrete appear to depend
not so much on the chemical composition of the hydrated cement as on the physical
structure of the products of hydration, viewed at the level of colloidal dimensions. For
this reason it is important to have a good picture of the physical properties of the
cement gel.
Fresh cement paste is a plastic network of particles of cement in water but, once
the paste has set, its apparent or gross volume remains approximately constant. At any
stage of hydration, the hardened paste consists of very poorly crystallized hydrates of
the various compounds, referred to collectively as gel, of crystals of Ca(OH)2, some
minor components, unhydrated cement, and the residue of the water-filled spaces in
the fresh paste. These voids are called capillary pores but, within the gel itself, there
exist interstitial voids, called gel pores. The nominal diameter of gel pores is about 3
nm while capillary pores are one or two orders of magnitude larger. There are thus, in
hydrated paste, two distinct classes of pores represented diagrammatically in Fig. 1.8.

Fig. 1.8. Simplified model of paste structure.1.22 Solid dots represent gel particles;
interstitial spaces are gel pores; spaces such as those marked C are capillary
pores. Size of gel pores is exaggerated
Because most of the products of hydration are colloidal (the ratio of calcium
silicate hydrates to Ca(OH)2 being 7 : 2 by mass1.60) during hydration the surface area
of the solid phase increases enormously, and a large amount of free water becomes
adsorbed on this surface. If no water movement to or from the cement paste is
permitted, the reactions of hydration use up the water until too little is left to saturate
the solid surfaces, and the relative humidity within the paste decreases. This is known
as self-desiccation. Because gel can form only in water-filled space, self-desiccation
leads to a lower hydration compared with a moist-cured paste. However, in self-
desiccated pastes with water/cement ratios in excess of 0.5, the amount of mixing
water is sufficient for hydration to proceed at the same rate as when moist-cured.
Volume of products of hydration
The gross space available for the products of hydration consists of the absolute
volume of the dry cement together with the volume of water added to the mix. The
small loss of water due to bleeding and the contraction of the paste while still plastic

31
will be ignored at this stage. The water bound chemically by C3S and C2S was shown
to be very approximately 24 and 21 per cent of the mass of the two silicates,
respectively. The corresponding figures for C3A and C4AF are 40 and 37 per cent,
respectively. The latter value is calculated on the assumption that the final reaction of
hydration of C4AF is, in approximate terms,

C4AF + 2Ca(OH)2 + 10H → C3AH6 + C3FH6.

As mentioned earlier, these figures are not accurate because our knowledge of
stoichiometry of the products of hydration of cement is inadequate to state the
amounts of water combined chemically. It is preferable, therefore, to consider non-
evaporable water as determined by a given method (see p. 36). This water, determined
under specified conditions,1.48 is taken as 23 per cent of the mass of anhydrous cement
(although in Type II cement the value may be as low as 18 per cent).
The specific gravity of the products of hydration of cement is such that they
occupy a greater volume than the absolute volume of unhydrated cement but smaller
than the sum of volumes of the dry cement and the non-evaporable water by
approximately 0.254 of the volume of the latter. An average value of specific gravity
of the products of hydration (including pores in the densest structure possible) in a
saturated state is 2.16.
As an example, let us consider the hydration of 100 g of cement. Taking the
specific gravity of dry cement as 3.15, the absolute volume of unhydrated cement is
100/3.15 = 31.8 ml. The non-evaporable water is, as we have said, about 23 per cent
of the mass of cement, i.e. 23 ml. The solid products of hydration occupy a volume
equal to the sum of volumes of anhydrous cement and water less 0.254 of the volume
of non-evaporable water, i.e.

31.8 + 0.23 × 100(1 – 0.254) = 48.9 ml.

Because the paste in this condition has a characteristic porosity of about 28 per cent,
the volume of gel water, wg, is given by

whence wg = 19.0 ml, and the volume of hydrated cement is 48.9 + 19.0 = 67.9 ml.
Summarizing, we have:

32
 It should be noted that the hydration was assumed to take place in a sealed test
tube with no water movement to or from the system. The volumetric changes are
shown in Fig. 1.9. The ‘decrease in volume’ of 5.9 ml represents the empty capillary
space distributed throughout the hydrated cement paste.

33
Fig. 1.9. Diagrammatic representation of volume changes on hydration of cement
paste with a water/cement ratio of 0.42
The values given above are only approximate but, had the total amount of water
been lower than about 42 ml, it would have been inadequate for full hydration as gel
can form only when sufficient water is available both for the chemical
reactions and for the filling of the gel pores being formed. The gel water, because it is
held firmly, cannot move into the capillaries so that it is not available for hydration of
the still unhydrated cement.
Thus, when hydration in a sealed specimen has progressed to a stage when the
combined water has become about one-half of the original water content, no further
hydration will take place. It follows also that full hydration in a sealed specimen is
possible only when the mixing water is at least twice the water required for chemical
reaction, i.e. the mix has a water/cement ratio of about 0.5 by mass. In practice, in the
example given above, the hydration would not in fact have progressed to completion
because hydration stops even before the capillaries have become empty. It has been

34
found that hydration becomes very slow when the water vapour pressure falls below
about 0.8 of the saturation pressure.1.23
Let us now consider the hydration of a cement paste cured under water so that
water can be imbibed as some of the capillaries become emptied by hydration. As
shown before, 100 g of cement (31.8 ml) will, on full hydration, occupy 67.9 ml. Thus,
for no unhydrated cement to be left and no capillary pores to be present, the original
mixing water should be approximately (67.9 – 31.8) = 36.1 ml. This corresponds to a
water/cement ratio of 1.14 by volume or 0.36 by mass. From other work, values of
about 1.2 and 0.38, respectively, have been suggested.1.22
If the actual water/cement ratio of the mix, allowing for bleeding, is less than about
0.38 by mass, complete hydration is not possible as the volume available is
insufficient to accommodate all the products of hydration. It will be recalled that
hydration can take place only in water within the capillaries. For instance, if we have
a mix of 100 g of cement (31.8 ml) and 30 g of water, the water would suffice to
hydrate x g of cement, given by the following calculations.
Contraction in volume on hydration is:

0.23x × 0.254 = 0.0585x.

Volume occupied by solid products of hydration is:

Porosity is:

and total water is 0.23x + wg = 30. Hence, x = 71.5 g = 22.7 ml and wg = 13.5 g. Thus,
the volume of hydrated cement is

0.489 × 71.5 + 13.5 = 48.5 ml.

The volume of unhydrated cement is 31.8 – 22.7 = 9.1 ml. Therefore, the volume of
empty capillaries is

(31.8 + 30) – (48.5 + 9.1) = 4.2 ml.

If water is available from outside, some further cement can hydrate, its quantity
being such that the products of hydration occupy 4.2 ml more than the volume of dry
cement. We found that 22.7 ml of cement hydrates to ocupy 48.5 ml, i.e. the products
of hydration of 1 ml of cement occupy 48.5/22.7 = 2.13 ml. Thus 4.2 ml would be
filled by the hydration of y ml of cement such that (4.2 + y)/y = 2.13; hence, y = 3.7
ml. Thus the volume of still unhydrated cement is 31.8 – (22.7 + 3.7) = 5.4 ml, which
has a mass of 17 g. In other words, 19 per cent of the original mass of cement has
remained unhydrated and can never hydrate because the gel already occupies all the
space available, i.e. the gel/space ratio (see p. 274) of the hydrated cement paste is 1.0.
It may be added that unhydrated cement is not detrimental to strength and, in fact,
among cement pastes all with a gel/space ratio of 1.0 those with a higher proportion of
unhydrated cement (i.e. a lower water/cement ratio) have a higher strength, possibly

35
because in such pastes the layers of hydrated paste surrounding the unhydrated
cement grains are thinner.1.24
Abrams obtained strengths of about 280 MPa (40 000 psi) using mixes with a
water/cement ratio of 0.08 by mass, but clearly considerable pressure is necessary to
obtain a properly consolidated mix of such proportions. Later on, Lawrence1.52 made
compacts of cement powder in a die assembly under a very high pressure (up to 672
MPa (or 97 500 psi)), using the techniques of powder metallurgy. Upon subsequent
hydration for 28 days, compressive strengths up to 375 MPa (or 54 500 psi) and
tensile strengths up to 25 MPa (or 3600 psi) were measured. The porosity of such
mixes and therefore the ‘equivalent’ water/cement ratio are very low. Even higher
strengths, up to 655 MPa (or 95 000 psi), were obtained using very high pressure and
a high temperature. The reaction products in these compacts were, however, different
from those resulting from normal hydration of cement.1.89
In contrast to these compacts which had an extremely low water/cement ratio, if
the water/cement ratio is higher than about 0.38 by mass, all the cement can hydrate
but capillary pores will also be present. Some of the capillaries will contain excess
water from the mix, others will fill by imbibing water from outside. Figure
1.10 shows the relative volumes of unhydrated cement, products of hydration, and
capillaries for mixes with different water/cement ratios.

36
 Fig. 1.10. Composition of cement paste at different stages of hydration.1.10 The
percentage indicated applies only to pastes with enough water-filled space to
accommodate the products at the degree of hydration indicated
As a more specific example, let us consider the hydration of a paste with a
water/cement ratio of 0.475, sealed in a tube. Let the mass of dry cement be 126 g,
which corresponds to 40 ml. The volume of water is then 0.475 × 126 = 60 ml. These
mix proportions are shown in the left-hand diagram of Fig. 1.11, but in reality the
cement and water are of course intermixed, the water forming a capillary system
between the unhydrated cement particles.

Fig. 1.11. Diagrammatic representation of the volumetric proportions of cement

paste at different stages of hydration
Let us now consider the situation when the cement has hydrated fully. The non-
evaporable water is 0.23 × 126 = 29.0 ml and the gel water is wg such that

37
whence the volume of gel water is 24.0 ml, and the volume of hydrated cement is 85.6
ml. There are thus 60 – (29.0 + 24.0) = 7.0 ml of water left as capillary water in the
paste. In addition, 100 – (85.6 + 7.0) = 7.4 ml form empty capillaries. If the cement
paste had access to water during curing these capillaries would fill with imbibed water.
This then is the situation at 100 per cent hydration when the gel/space ratio is
0.856, as shown in the right-hand diagram of Fig. 1.11. As a further illustration, the
centre diagram shows the volumes of different components when only half the cement
has hydrated. The gel/space ratio is then

An approach similar to that of Powers, outlined above, has been applied to cements
containing a limestone filler (see p. 88).1.97
Capillary pores
We can thus see that, at any stage of hydration, the capillary pores represent that part
of the gross volume which has not been filled by the products of hydration. Because
these products occupy more than twice the volume of the original solid phase (i.e.
cement) alone, the volume of the capillary system is reduced with the progress of
hydration.
Thus the capillary porosity of the paste depends both on the water/cement ratio of
the mix and on the degree of hydration. The rate of hydration of the cement is of no
importance per se, but the type of cement influences the degree of hydration achieved
at a given age. As mentioned before, at water/cement ratios higher than about 0.38,
the volume of the gel is not sufficient to fill all the space available to it so that there
will be some volume of capillary pores left even after the process of hydration has
been completed.
Capillary pores cannot be viewed directly but their median size was estimated
from vapour pressure measurement to be about 1.3 μm. In fact, the size of pores in
hydrated cement paste varies widely. Glasser’s studies1.85 indicate that mature cement
pastes contain few pores larger than 1 μm, with most pores being smaller than 100 nm.
They vary in shape but, as shown by measurement of permeability, form an
interconnected system randomly distributed throughout the cement paste.125 These
interconnected capillary pores are mainly responsible for the permeability of the
hardened cement paste and for its vulnerability to cycles of freezing and thawing.
However, hydration increases the solid content of the paste and, in mature and
dense pastes, the capillaries can become blocked by gel and segmented so that they
turn into capillary pores interconnected solely by the gel pores. The absence of
continuous capillaries is due to a combination of a suitable water/cement ratio and a
sufficiently long period of moist curing; the degree of maturity required for different
water/cement ratios for ordinary Portland cements is indicated in Fig. 1.12. The actual
time to achieve the required maturity depends on the characteristics of the cement
used, but approximate values of the time required can be gauged from the data
of Table 1.6. For water/cement ratios above about 0.7, even complete hydration would
not produce enough gel to block all the capillaries. For extremely fine cement, the
maximum water/cement ratio would be higher, possibly up to 1.0; conversely, for
coarse cements, it would be below 0.7. The importance of eliminating continuous
capillaries is such that this might be regarded a necessary condition for a concrete to
be classified as ‘good’.

38
Fig. 1.12. Relation between the water/cement ratio and the degree of hydration at
which the capillaries cease to be continuous1.26

Table 1.6. Approximate Age Required to Produce Maturity at which Capillaries

Become Segmented1.26

Gel pores
Let us now consider the gel itself. From the fact that it can hold large quantities of
evaporable water it follows that the gel is porous, but the gel pores are
really interconnected interstitial spaces between the gel particles, which are needle-,
plate-, and foil-shaped. The gel pores are much smaller than the capillary pores: less
than 2 or 3 nm in nominal diameter. This is only one order of magnitude greater than
the size of molecules of water. For this reason, the vapour pressure and mobility of
adsorbed water are different from the corresponding properties of free water. The
amount of reversible water indicates directly the porosity of the gel.1.24

39
 The gel pores occupy about 28 per cent of the total volume of gel, the material left
after drying in a standard manner1.48 being considered as solids. The actual value is
characteristic for a given cement but is largely independent of the water/cement ratio
of the mix and of the progress of hydration. This would indicate that gel of similar
properties is formed at all stages and that continued hydration does not affect the
products already in existence. Thus, as the total volume of gel increases with the
progress of hydration, the total volume of gel pores also increases. On the other hand,
as mentioned earlier, the volume of capillary pores decreases with the progress of
hydration.
Porosity of 28 per cent means that the gel pores occupy a space equal to about one-
third of the volume of the gel solids. The ratio of the surface of the solid part of the
gel to the volume of the solids is equal to that of spheres about 9 nm in diameter. This
must not be construed to mean that gel consists of spherical elements; the solid
particles are of varied shapes, and bundles of such particles form a cross-linked
network containing some more or less amorphous interstitial material.1.27
Another way of expressing the porosity of the gel is to say that the volume of the
pores is about three times the volume of the water forming a layer one molecule thick
over the entire solid surface in the gel.
From measurements of water adsorption, the specific surface of gel has been
estimated to be of the order of 5.5 × 108 m2 per m3, or approximately 200 000
m2/kg.1.27 Small-angle X-ray scattering measurements have yielded values of the order
of 600 000 m2/kg, indicating a large internal surface within the particles.1.63 By contrast,
unhydrated cement has a specific surface of some 200 to 500 m2/kg. At the other
extreme, silica fume has a specific surface of 22 000 m2/kg.
In connection with the pore structure, it may be relevant to note that high-pressure
steam-cured cement paste has a specific surface of some 7000 m2/kg only. This
indicates an entirely different particle size of the products of hydration at a high
pressure and a high temperature and, in fact, such treatment results in an almost
entirely micro-crystalline material rather than gel.
The specific surface of normally cured cement paste depends on the curing
temperature and on the chemical composition of cement. It has been suggested1.27 that
the ratio of the specific surface to the mass of non-evaporable water (which in turn is
proportional to the porosity of the hydrated cement paste) is proportional to

0.230(C3S) + 0.320(C2S) + 0.317(C3A) + 0.368(C4AF),

where the symbols in brackets represent the percentages of the compounds present in
the cement. There seems to be little variation between the numerical coefficients of
the last three compounds, and this indicates that the specific surface of the hydrated
cement paste varies little with a change in the composition of cement. The rather
lower coefficient of C3S is due to the fact that it produces a large quantity of micro-
crystalline Ca(OH)2, which has a very much lower specific surface than the gel.
The proportionality between the mass of water forming a monomolecular layer
over the surface of the gel and the mass of non-evaporable water in the paste (for a
given cement) means that gel of nearly the same specific surface is formed throughout
the progress of hydration. In other words, particles of the same size are formed all the
time and the already existing gel particles do not grow in size. This is not, however,
the case in cements with a high C2S content.1.28

40
Mechanical strength of cement gel
There are two classical theories of hardening or development of strength of cement.
That put forward by H. Le Chatelier in 1882 states that the products of hydration of
cement have a lower solubility than the original compounds, so that the hydrates
precipitate from a supersaturated solution. The precipitate is in the form of interlaced
elongated crystals with high cohesive and adhesive properties.
The colloidal theory propounded by W. Michaëlis in 1893 states that the
crystalline aluminate, sulfoaluminate and hydroxide of calcium give the initial
strength. The lime-saturated water then attacks the silicates and forms a hydrated
calcium silicate which, being almost insoluble, forms a gelatinous mass. This mass
hardens gradually due to the loss of water either by external drying or by hydration of
the inner unhydrated core of the cement grains: in this manner cohesion is obtained.
In the light of modern knowledge it appears that both theories contain elements of
truth and are in fact by no means irreconcilable. In particular, colloid chemists have
found that many, if not most, colloids consist of crystalline particles but these, being
extremely small, have a large surface area which gives them what appear to be
different properties from the usual solids. Thus colloidal behaviour is essentially a
function of the size of the surface area rather than of the non-regularity of internal
structure of the particles involved.1.42
In the case of Portland cement, it has been found that, when mixed with a large
quantity of water, cement produces within a few hours a solution supersaturated with
Ca(OH)2 and containing concentrations of calcium silicate hydrate in a metastable
condition.1.2 This hydrate rapidly precipitates in agreement with Le Chatelier’s
hypothesis; the subsequent hardening may be due to the withdrawal of water from the
hydrated material as postulated by Michaëlis. Following the dormant period,
precipitation of calcium silicate hydrate and Ca(OH)2 continues.
Further experimental work has shown that the calcium silicate hydrates are in fact
in the form of extremely small (measured in nanometres) interlocking
crystals1.20 which, because of their size, could equally well be described as gel. When
cement is mixed with a small quantity of water, the degree of crystallization is
probably even poorer, the crystals being ill-formed. Thus the Le Chatelier–Michaëlis
controversy is largely reduced to a matter of terminology as we are dealing with a gel
consisting of crystals. Moreover, the solubility of silica increases very substantially at
a pH above 10, so that it is possible for the Michaëlis mechanism to operate initially
and for that of Le Chatelier later on. A more detailed discussion of the two
mechanisms is offered by Baron and Santeray.1.94
The term ‘cement gel’ is considered, for convenience, albeit not correctly, to
include the crystalline calcium hydroxide. Gel is thus taken to mean the cohesive
mass of hydrated cement in its densest paste, i.e. inclusive of gel pores, the
characteristic porosity being about 28 per cent.
The actual source of strength of the gel is not fully understood but it probably
arises from two kinds of cohesive bonds.1.27 The first type is the physical attraction
between solid surfaces, separated only by the small (less than 3 nm) gel pores; this
attraction is usually referred to as van der Waals’ forces.
The source of the second type of cohesion is the chemical bonds. Because cement
gel is of the limited swelling type (i.e. the particles cannot be dispersed by addition of
water) it seems that the gel particles are cross-linked by chemical forces. These are
much stronger than van der Waals’ forces but the chemical bonds cover only a small

41
fraction of the boundary of the gel particles. On the other hand, a surface area as high
as that of cement gel is not a necessary condition for high strength development, as
high-pressure steam-cured cement paste, which has a low surface area, exhibits
extremely good hydraulic properties.1.14
We cannot thus estimate the relative importance of the physical and chemical
bonds but there is no doubt that both contribute to the very considerable strength of
the hardened cement paste. It has to be admitted that the understanding of the
cohesive nature of the hydrated cement paste and its adhesion to aggregate is still
imperfect. As Nonat and Mutin1.92 put it, the microstructure has not been related in a
general way to mechanical properties.
Water held in hydrated cement paste
The presence of water in hydrated cement has been repeatedly mentioned. The cement
paste is indeed hygroscopic owing to the hydrophilic character of cement coupled
with the presence of sub-microscopic pores. The actual water content of the paste
depends on the ambient humidity. In particular, capillary pores, because of their
comparatively large size, empty when the ambient relative humidity falls below about
45 per cent,1.25 but water is adsorbed in the gel pores even at very low ambient
humidities.
We can thus see that water in hydrated cement is held with varying degrees of
firmness. At one extreme, there is free water; at the other, chemically combined water
forming a definite part of the hydrated compounds. Between these two categories,
there is gel water held in a variety of other ways.
The water held by the surface forces of the gel particles is called adsorbed water,
and that part of it which is held between the surfaces of certain planes in a crystal is
called interlayer or zeolitic water. Lattice water is that part of the water of
crystallization which is not chemically associated with the principal constituents of
the lattice. The diagrammatic representation of Fig. 1.13 may be of interest.

Fig. 1.13. Probable structure of hydrated silicates1.53

Free water is held in capillaries and is beyond the range of the surface forces of the
solid phase.
There is no technique available for determining how water is distributed between
these different states, nor is it easy to predict these divisions from theoretical
considerations as the energy of binding of combined water in the hydrate is of the
same order of magnitude as the energy of binding of the adsorbed water. However,

42
investigations using nuclear magnetic resonance suggest that gel water has the same
energy of binding as interlayer water in some swelling clays; thus the gel water may
well be in interlayer form.1.54
A convenient division of water in the hydrated cement, necessary for investigation
purposes, though rather arbitrary, is into two categories: evaporable and non-
evaporable. This is achieved by drying the cement paste to equilibrium (i.e. to a
constant mass) at a given vapour pressure. The usual value is 1 Pa at 23 °C, obtained
over Mg(ClO4)2.2H2O. Drying in an evacuated space which is connected to a moisture
trap held at –79 °C has also been used; this corresponds to a vapour pressure of 0.07
Pa.1.48 Alternatively, the evaporable water can be determined by the loss upon drying at
a higher temperature, usually 105 °C, or by freezing out, or by removing with a
solvent.
All these methods essentially divide water according to whether or not it can be
removed at a certain reduced vapour pressure. Such a division is perforce arbitrary
because the relation between vapour pressure and water content of cement is
continuous; in contrast to crystalline hydrates, no breaks occur in this relationship.
However, in general terms, the non-evaporable water contains nearly all chemically
combined water and also some water not held by chemical bonds. This water has a
vapour pressure lower than that of the ambient atmosphere and the quantity of such
water is, in fact, a continuous function of the ambient vapour pressure.
The amount of non-evaporable water increases as hydration proceeds but, in a
saturated paste, non-evaporable water can never become more than one-half of the
total water present. In well-hydrated cement, the non-evaporable water is about 18 per
cent by mass of the anhydrous material; this proportion rises to about 23 per cent in
fully hydrated cement.1.1 It follows from the proportionality between the amount of
non-evaporable water and the solid volume of the cement paste that the former
volume can be used as a measure of the quantity of the cement gel present, i.e. of the
degree of hydration.
The manner in which water is held in a cement paste determines the energy of
binding. For instance, 1670 J (400 calories) are used in establishing the bond of 1 g of
non-evaporable water, while the energy of the water of crystallization of Ca(OH)2 is
3560 J/g (850 cal/g). Likewise, the density of the water varies; it is approximately 1.2
for non-evaporable water, 1.1 for gel water, and 1.0 for free water.1.24 It has been
suggested that the increase in the density of the adsorbed water at low surface
concentrations is not the result of compression but is caused by the orientation, or
ordering, of the molecules in the adsorbed phase due to the action of the surface
forces,1.12 resulting in a so-called disjoining pressure. The disjoining pressure is the
pressure expected to maintain the film of adsorbed molecules against external action.
A confirmation of the hypothesis that the properties of adsorbed water are different
from those of free water is afforded by measurements of the absorption of
microwaves by hardened cement paste.1.64
Heat of hydration of cement
In common with many chemical reactions, the hydration of cement compounds is
exothermic, energy of up to 500 J/g (120 cal/g) of cement being liberated. Because the
thermal conductivity of concrete is comparatively low, it acts as an insulator, and in
the interior of a large concrete mass, hydration can result in a large rise in temperature.
At the same time, the exterior of the concrete mass loses some heat so that a steep
temperature gradient may be established and, during subsequent cooling of the

43
interior, serious cracking may result. This behaviour is, however, modified by the
creep of concrete or by insulation of the surfaces of the concrete mass.
At the other extreme, the heat produced by the hydration of cement may prevent
freezing of the water in the capillaries of freshly placed concrete in cold weather, and
a high evolution of heat is therefore advantageous. It is clear, then, that it is advisable
to know the heat-producing properties of different cements in order to choose the
most suitable cement for a given purpose. It may be added that the temperature of
young concrete can also be influenced by artificial heating or cooling.
The heat of hydration is the quantity of heat, in joules per gram of unhydrated
cement, evolved upon complete hydration at a given temperature. The most common
method of determining the heat of hydration is by measuring the heats of solution of
unhydrated and hydrated cement in a mixture of nitric and hydrofluoric acids: the
difference between the two values represents the heat of hydration. This method is
described in BS 4550-3.8 : 1978, and is similar to the method of ASTM C 186-05.
While there are no particular difficulties in this test, care should be taken to prevent
carbonation of the unhydrated cement because the absorption of 1 per cent of
CO2 results in an apparent decrease in the heat of hydration of 24.3 J/g (5.8 cal/g) out
of a total of between 250 and over 420 J/g (60 and 100 cal/g).1.29
The temperature at which hydration takes place greatly affects the rate of heat
development, as shown by the data of Table 1.7, which gives the heat developed in 72
hours at different temperatures.1.30 There is little effect of the temperature on the long-
term value of the heat of hydration.1.82

Table 1.7. Heat of Hydration Developed After 72 Hours at Different

Temperatures1.30

Strictly speaking, the heat of hydration, as measured, consists of the chemical heat
of the reactions of hydration and the heat of adsorption of water on the surface of the
gel formed by the processes of hydration. The latter heat accounts for about a quarter
of the total heat of hydration. Thus, the heat of hydration is really a composite
quantity.1.24
For practical purposes, it is not necessarily the total heat of hydration that matters
but the rate of heat evolution. The same total heat produced over a longer period can
be dissipated to a greater degree with a consequent smaller rise in temperature. The
rate of heat development can be easily measured in an adiabatic calorimeter, and
typical time–temperature curves obtained under adiabatic conditions are shown in Fig.
1.14. (The ratio 1:2:4 represents the proportion by mass of cement:fine
aggregate:coarse aggregate.)

44
 Fig. 1.14. Temperature rise in 1:2:4 concrete (water/cement ratio of 0.60) made
with different cements and cured adiabatically.1.31 The total heat of hydration of
each cement at three days is shown (Crown copyright)
For the usual range of Portland cements, Bogue1.2 observed that about one-half of
the total heat is evolved between 1 and 3 days, about three-quarters in 7 days, and 83
to 91 per cent of the total heat in 6 months. The actual value of the heat of hydration
depends on the chemical composition of the cement, and is very nearly a sum of the
heats of hydration of the individual compounds when hydrated separately. It follows
that, given the compound composition of a cement, its heat of hydration can be
calculated with a fair degree of accuracy. Typical values of the heat of hydration of
pure compounds are given in Table 1.8.

Table 1.8. Heat of Hydration of Pure Compounds1.32

It may be noted that there is no relation between the heat of hydration and the
cementing properties of the individual compounds. Woods, Steinour
and Starke1.33 tested a number of commercial cements and, using the method of least
squares, calculated the contribution of individual compounds to the total heat of
hydration of cement. They obtained equations of the following type: heat of hydration
of 1 g of cement is

136(C3S) + 62(C2S) + 200(C3A) + 30(C4AF)

45
where the terms in brackets denote the percentage by mass of the individual
compounds in cement. Later work1.83 broadly confirmed the contribution of the various
compounds to the heat of hydration of cement except for C2S whose contribution was
found to be about one-half of that given above.
Because in the early stages of hydration the different compounds hydrate at
different rates, the rate of heat evolution, as well as the total heat, depends on the
compound composition of the cement. It follows that by reducing the proportions of
the compounds that hydrate most rapidly (C3A and C3S) the high rate of heat evolution
in the early life of concrete can be lowered. The fineness of the cement also influences
the rate of heat development, an increase in fineness speeding up the reactions of
hydration and therefore the heat evolved. It is reasonable to assume that the early rate
of hydration of each compound in cement is proportional to the surface area of the
cement. However, at later stages, the effect of the surface area is negligible and the
total amount of heat evolved is not affected by the fineness of cement.
The influence of C3A and C3S can be gauged from Figs 1.15 and 1.16. As
mentioned before, for many uses of concrete, a controlled heat evolution is
advantageous and suitable cements have been developed. One such cement is low
heat Portland cement discussed in more detail in Chapter 2. The rate of heat
development of this and other cements is shown in Fig. 1.17.

Fig. 1.15. Influence of C3A content on heat evolution1.32 (C3S content

approximately constant)

46
 Fig. 1.16. Influence of C3S content on heat evolution1.32 (C3A content
approximately constant)

47
 Fig. 1.17. Development of heat of hydration of different cements cured at 21 °C
(70 °F) (water/cement ratio of 0.40)1.34
The quantity of cement in the mix also affects the total heat development: thus the
richness of the mix, that is, the cement content, can be varied in order to help the
control of heat development.
Influence of the compound composition on properties of cement
In the preceding section, it was shown that the heat of hydration of cement is a simple
additive function of the compound composition of cement. It would seem, therefore,
that the various hydrates retain their identity in the cement gel, which can be
considered thus to be a fine physical mixture or to consist of copolymers of the
hydrates. A further corroboration of this supposition is obtained from the
measurement of specific surface of hydrated cements containing different amounts of
C3S and C2S: the results agree with the specific surface areas of hydrated neat C3S and
C2S. Likewise, the water of hydration agrees with the additivity of the individual
compounds.
This argument does not, however, extend to all properties of hardened cement
paste, notably to shrinkage, creep, and strength; nevertheless, the compound
composition gives some indication of the properties to be expected. In particular, the
composition controls the rate of evolution of the heat of hydration and the resistance
of cement to sulfate attack, so that limiting values of oxide or compound composition
of different types of cement are prescribed by some specifications. The limitations of
ASTM C 150-09 are less restrictive than they used to be (see Table 1.9).

Table 1.9. Compound Composition Limits for Cements of ASTM C 150-09

The difference in the early rates of hydration of C3S and C2S – the two silicates
primarily responsible for the strength of hydrated cement paste – has been mentioned
earlier. A convenient approximate rule assumes that C3S contributes most to the
strength development during the first four weeks and C2S influences the gain in
strength from 4 weeks onwards.1.35 At the age of about one year, the two compounds,
mass for mass, contribute approximately equally to the ultimate strength.1.36 Pure C3S
and C2S have been found to have a strength of about 70 MPa (10 000 psi) at the age of
18 months, but at the age of 7 days C2S had no strength while the strength of C3S was
about 40 MPa (6000 psi). The usually accepted development of strength of pure
compounds is shown in Fig. 1.18.

48
 Fig. 1.18. Development of strength of pure compounds according to Bogue1.2
However, these relative values of the contribution to strength of the individual
compounds in Portland cement have been challenged.1.87 Tests, using particles with the
same size distribution and at a fixed water/solid ratio of 0.45, have shown that, up to
at least the age of 1 year, C2S exhibits a lower strength than C3S. Nevertheless, both
silicates are much stronger than C3A and C4AF, although the latter compound exhibits
a significant strength while C3A has a negligible strength1.87 (see Fig. 1.19).

49
 Fig. 1.19. Development of strength of pure compounds according to Beaudoin
and Ramachandran (reprinted from ref. 1.87 by kind permission of Elsevier
Science Ltd, Kidlington, U.K.)
As mentioned on p. 14, the calcium silicates appear in commercial cements in
‘impure’ form. These impurities may strongly affect the rate of reaction and of
strength development of the hydrates. For instance, the addition of 1 per cent of
A12O3 to pure C3S increases the early strength of the hydrated paste, as shown in Fig.
1.20. According to Verbeck,1.55 this increase in strength probably results from
activation of the silicate crystal lattice due to introduction of the alumina (or magnesia)
into the crystal lattice with resultant activating structural distortions.

Fig. 1.20. Development of strength of pure C3S and C3S with 1 per cent of Al2O31.55
The rate of hydration of C2S is also accelerated by the presence of other
compounds in cement but, within the usual range of the C2S content in modern
Portland cements (up to 30 per cent) the effect is not large.
The influence of the other major compounds on the strength development of
cement has been established less clearly. C3A contributes to the strength of the cement
paste at one to three days, and possibly longer, but causes retrogression at an
advanced age, particularly in cements with a high C3A or (C3A + C4AF) content. The
role of C3A is still controversial, but is not important with respect to strength in
practice.
The role of C4AF in the development of strength of cement is also debatable, but
there certainly is no appreciable positive contribution. It is likely that colloidal
hydrated CaO.Fe2O3 is deposited on the cement grains, thus delaying the progress of
hydration of other compounds.1.7
From the knowledge of the contribution to strength of the individual compounds
present, it might be thought possible to predict the strength of cement on the basis of
its compound composition. This would be in the form of an expression of the type:

strength = a(C3S) + b(C2S) + c(C3A) + d(C4AF),

where the symbols in brackets represent the percentage by mass of the compound,
and a, b, etc. are constants representing the contribution of 1 per cent of the
corresponding compound to the strength of the hydrated cement paste.

50
 The use of such an expression would make it easy to forecast, at the time of
manufacture, the strength of cement and would reduce the need for conventional
testing. Such a relation does indeed exist in laboratory tests using cements prepared
from the pure four main compounds. In practice, however, the contribution of
different compounds is not simply additive and has been found to depend on age and
on the curing conditions.
All that can be said is that, in general terms, an increase in the C3S content
increases strength up to 28 days;1.56 Figure 1.21 shows the 7-day strength of standard
mortars made with cements of different composition and obtained from different
plants.1.37 The C2S content has a positive influence on strength at 5 and 10 years only,
and C3A a positive influence up to 7 or 28 days but a negative influence later
on.1.56,1.57 The influence of the alkalis is considered on p. 46. Prediction of the effects of
compounds other than silicates on strength is unreliable According to Lea,1.38 these
discrepancies may be due to the presence of glass in clinker, discussed more fully in
the succeeding section.

Fig. 1.21. Relation between 7-day strength of cement paste and the C3S content in
cement.1.37 Each mark represents cement from one plant
An extensive review by Odler1.79 has shown, moreover, that a generally applicable
strength prediction equation for commercial cements is not possible for several
reasons. These are: the interaction between the compounds; the influence of the
alkalis and of gypsum; and the influence of the particle size distribution of the cement.
The presence of glass, which does not contain all the compounds in the same
proportions as the rest of the clinker, but which affects reactivity, as well as the
amount of free lime, are also factors varying between cements with nominally the
same composition of the four main compounds.
Attempts1.93 have been made to generate strength prediction equations for mortar on
the basis of parameters which include, in addition to the main compound composition,
terms for SO3, CaO, MgO and the water/cement ratio, but the reliability of prediction
is marginal.
From the foregoing, we can conclude that the relations between strength and
compound composition of Portland cements in general which have been observed are
stochastic in nature. Deviations from these relations arise from the fact that they

51
ignore some of the variables involved.1.14 It can be argued, in any case, that all
constituents of hydrated Portland cement contribute in some measure to strength in so
far as all products of hydration fill space and thus reduce porosity.
Furthermore, there are some indications that the additive behaviour cannot be fully
realized. In particular, Powers1.22 suggested that the same products are formed at all
stages of hydration of the cement paste; this follows from the fact that, for a given
cement, the surface area of hydrated cement is proportional to the amount of water of
hydration, whatever the water/cement ratio and age. Thus the fractional rates of
hydration of all compounds in a given cement would be the same. This is probably the
case only after the rate of diffusion through the gel coating has become the rate-
determining factor, but not at early ages,1.65 say up to 7 days.1.49 Confirmation of equal
fractional rate of hydration was obtained by Khalil and Ward,1.70 but we now accept
that early hydration of the different compounds proceeds at different rates; later on,
the rates become equal.
There is another factor influencing the rate of hydration: the fact that the
composition is not the same at different points in space. This arises from the fact that,
for diffusion to take place from the face of the still unhydrated part of the cement
grain to the space outside (see p. 13), there must be a difference in ion concentration:
the space outside is saturated but that inside is supersaturated. This diffusion varies
the rate of hydration.
It is likely, therefore, that neither the suggestion of equal fractional rates of
hydration, nor the assumption that each compound hydrates at a rate independent of
other compounds, is valid. Indeed, we have to admit that our understanding of the
hydration rates is still unsatisfactory.
For instance, the amount of heat of hydration per unit mass of hydrated material
has been found to be constant at all ages1.34 (see Fig. 1.22), thus suggesting that the
nature of the products of hydration does not vary with time. It is therefore reasonable
to use the assumption of equal fractional rates of hydration within the limited range of
composition of ordinary and rapid-hardening Portland cements. However, other
cements which have a higher C2S content than ordinary cement or rapid-hardening
cement do not conform to this behaviour. Measurements of heat of hydration indicate
that C3S hydrates earlier, and some C2S is left to hydrate later.

52
 Fig. 1.22. Relation between the heat of hydration and the amount of non-
evaporable water for ordinary Portland cement1.22
Furthermore, the initial framework of the paste established at the time of setting
affects to a large degree the subsequent structure of the products of hydration. This
framework influences especially the shrinkage and development of strength.1.14 It is not
surprising, therefore, that there is a definite relation between the degree of hydration
and strength. Figure 1.23 shows, for instance, an experimental relation between the
compressive strength of concrete and the combined water in a cement paste with a
water/cement ratio of 0.25.1.39 These data agree with Powers’ observations on the
gel/space ratio, according to which the increase in strength of a cement paste is a
function of the increase in the relative volume of gel, regardless of age, water/cement
ratio, or compound composition of cement. However, the total surface area of the
solid phase is related to the compound composition, which does affect the actual value
of the ultimate strength.1.22

53
 Fig. 1.23. Relation between compressive strength and combined water content1.1
Effects of alkalis
The effects of the minor compounds on the strength of cement paste are complex and
not yet fully established. Tests1.3 on the influence of alkalis have shown that the
increase in strength beyond the age of 28 days is strongly affected by the alkali
content: the greater the amount of alkali present the lower the gain in strength. This
has been confirmed by two statistical evaluations of strength of several hundred
commercial cements.1.56,1.57 The poor gain in strength between 3 and 28 days can be
attributed more specifically to water-soluble K2O present in the cement.1.58 On the
other hand, in the total absence of alkalis, the early strength of cement paste can be
abnormally low.1.58 Accelerated strength tests (see p. 623) have shown that, up to 0.4
per cent of Na2O, strength increases with an increase in the alkali content1.75 (Fig. 1.24).

54
 Fig. 1.24. Effect of alkali content on accelerated strength1.75
The influence of the alkalis on strength is complicated by the fact that they may be
incorporated into the calcium silicate hydrates or may exist as soluble sulfates; their
action in the two cases is not the same. K2O is believed to replace one molecule of
CaO in C2S with a consequent rise in the C3S content above that calculated.1.6 However,
we can say that, generally, the alkalis increase the early strength development and
reduce the long-term strength.1.79 Osbæck1.95 confirmed that a higher alkali content in
Portland cement increases the early strength and decreases the long-term strength.
The alkalis are known to react with the so-called alkali-reactive aggregates (see
p. 144), and cements used under such circumstances often have their alkali content
limited to 0.6 per cent (measured as equivalent Na2O). Such cements are referred to
as low-alkali cements.
One other consequence of the presence of alkalis in cement should be mentioned.
Fresh Portland cement paste has a very high alkalinity (pH above 12.5) but, in a
cement with a high alkali content, the pH is even higher. In consequence, human skin
is attacked and dermatitis or burns may result; eyes can also be injured. For this
reason, the use of protective clothing is essential.
We can see then that the alkalis are an important constituent of cement, but full
information on their role is yet to be obtained. It may be noted that the use of pre-
heaters in modern dry-process cement plants has led to an increase in the alkali
content of cement made from given raw materials. The alkali content, therefore, has to
be controlled, but limiting the alkali content too severely results in an increased
energy consumption.1,76 A more efficient dust collection also increases the alkali
content of the cement when the dust is re-incorporated into the cement because the
dust contains a large amount of alkalis; this may be as high as 15 per cent, in which
case the dust, or some of it, has to be discarded.
Effects of glass in clinker
It may be recalled that, during the formation of cement clinker in the kiln, some 20 to
30 per cent of the material becomes liquid; on subsequent cooling, crystallization
takes place but there is always some material which undercools to glass. In fact, the
rate of cooling of clinker greatly affects the properties of cement: if cooling were so

55
slow that full crystallization could be achieved (e.g. in a laboratory), β-C2S might
become converted to γ-C2S, this conversion being accompanied by expansion and
powdering, known as dusting. Furthermore, γ-C2S hydrates too slowly to be a useful
cementitious material. However, Al2O3, MgO and the alkalis may stabilize β-C2S, even
on very slow cooling in all practical cases.
Another reason why some glass is desirable is the effect of glass on the crystalline
phases. Alumina and ferric oxide are completely liquefied at clinkering temperatures,
and on cooling produce C3A and C4AF. The extent of glass formation would thus
affect these compounds to a large degree while the silicates, which are formed mainly
as solids, would be relatively unaffected. It may be noted, too, that glass can also hold
a large proportion of minor compounds such as the alkalis and MgO; the latter is thus
not available for expansive hydration.1.40 It follows that rapid cooling of high-magnesia
clinkers is advantageous. Because the aluminates are attacked by sulfates, their
presence in glass would also be an advantage. C3A and C4AF in glass form can
hydrate to a solid solution of C3AH6 and C3FH6 which is resistant to sulfates. However,
a high glass content adversely affects the grindability of clinker.
On the other hand, there are some advantages of a lower glass content. In some
cements, a greater degree of crystallization leads to an increase in the amount of C3S
produced.
It can be seen, then, that a strict control of the rate of cooling of clinker so as to
produce a desired degree of crystallization is very important. The range of glass
content in commercial clinkers, determined by the heat of solution method, is between
2 and 21 per cent.1.41 An optical microscope indicates much lower values.
It may be recalled that the Bogue compound composition assumes that the clinker
has crystallized completely to yield its equilibrium products, and, as we have seen, the
reactivity of glass is different from that of crystals of similar composition.
It can be seen then that the rate of cooling of clinker, as well as, possibly, other
characteristics of the process of cement manufacture, affect the strength of cement,
and this defies attempts to develop an expression for strength as a function of cement
composition. Nevertheless, if one process of manufacture is used and the rate of
cooling of clinker is kept constant, there is a definite relation between compound
composition and strength.
Tests on properties of cement
The manufacture of cement requires stringent control, and a number of tests are
performed in the cement plant laboratory to ensure that the cement is of the desired
quality and that it conforms to the requirements of the relevant national standards. It is
desirable, nonetheless, for the purchaser or for an independent laboratory to make
acceptance tests or, more frequently, to examine the properties of a cement to be used
for some special purpose. Tests on the chemical composition and fineness are
prescribed in European Standards BS EN 196-1 : 2005 and BS EN 196-6 : 2010,
respectively; further tests are prescribed by BS 4550-3-1 : 1978 for ordinary and
rapid-hardening Portland cements. Other relevant standards are mentioned when other
types of cement are discussed in Chapter 2.
Consistency of standard paste
For the determination of the initial and final setting times and for the Le Chatelier
soundness test, neat cement paste of a standard consistency has to be used. It is,

56
therefore, necessary to determine for any given cement the water content of the paste
which will produce the desired consistency.
The consistency is measured by the Vicat apparatus shown in Fig. 1.25, using a 10
mm diameter plunger fitted into the needle holder. A trial paste of cement and water is
mixed in a prescribed manner and placed in the mould. The plunger is then brought
into contact with the top surface of the paste and released. Under the action of its
weight the plunger will penetrate the paste, the depth of penetration depending on the
consistency. This is considered to be standard, in the meaning of BS EN 196-3 : 2005,
when the plunger penetrates the paste to a point 6 ± 1 mm from the bottom of the
mould. The water content of the standard paste is expressed as a percentage by mass
of the dry cement, the usual range of values being between 26 and 33 per cent.

Fig. 1.25. Vicat apparatus

Setting time
The physical processes of setting were discussed on p. 19; here, the actual
determination of setting times will be briefly dealt with. The setting times of cement
are measured using the Vicat apparatus (Fig. 1.25) with different penetrating
attachments. The test method is prescribed by BS EN 196-3 : 2005.
For the determination of the initial set, a round needle with a diameter 1.13 ± 0.05
mm is used. This needle, acting under a prescribed weight, is used to penetrate a paste
of standard consistency placed in a special mould. When the paste stiffens sufficiently
for the needle to penetrate no deeper than to a point 5 ± 1 mm from the bottom, initial
set is said to have taken place. Initial set is expressed as the time elapsed since the
mixing water was added to the cement. A minimum time of 60 minutes is prescribed
by BS EN 197-1 : 2000 for cements with strengths of 42.5 MPa, 75 minutes for
cements with a strength of 52.5 MPa and 45 minutes for cements with higher
strengths. American Standard ASTM C 150-09 prescribes a minimum time for the

57
initial set of 45 minutes, also using the Vicat apparatus prescribed in ASTM C 191-08.
An alternative test using Gillmore needles (ASTM C 266-08) gives a higher value of
setting time.
The initial setting time of high-alumina cement is prescribed by BS 915-2 : 1972
(1983) as between 2 and 6 hours.
Final set is determined by a similar needle fitted with a metal attachment hollowed
out so as to leave a circular cutting edge 5 mm in diameter and set 0.5 mm behind the
tip of the needle. Final set is said to have taken place when the needle, gently lowered
to the surface of the paste, penetrates it to a depth of 0.5 mm but the circular cutting
edge fails to make an impression on the surface of the paste. The final setting is
reckoned from the moment when mixing water was added to the cement. Limits on
the final setting time no longer appear in the European or ASTM standards.
If the knowledge of final setting time is required, but no test data are available, it
may be useful to take advantage of the observation that, for the majority of American
commercial ordinary and rapid-hardening Portland cements at room temperature, the
initial and final setting times are approximately related as follows: final setting time
(min) = 90 + 1.2 × initial setting time (min).
Because the setting of cement is affected by the temperature and the humidity of
the surrounding air, these are specified by BS EN 196-3 : 2005: a temperature of 20 ±
2 °C (68 ± 4 °F) and minimum relative humidity of 65 per cent.
Tests1.59 have shown that setting of cement paste is accompanied by a change in the
ultrasonic pulse velocity through it (cf. p. 633) but it has not been possible to develop
an alternative method of measurement of setting time of cement. Attempts at using
electrical measurements have also been unsuccessful, mainly because of the influence
of admixtures on electrical properties.1.73
It should be remembered that the speed of setting and the rapidity of hardening, i.e.
of gain of strength, are independent of one another. For instance, the prescribed
setting times of rapid-hardening cement are no different from those for ordinary
Portland cement, although the two cements harden at different rates.
It may be relevant to mention here that the setting time of concrete can also be
determined, but this is a different property from the setting time of cement. ASTM
Standard C 403-08 lays down the procedure for the former, which uses a Proctor
penetration probe applied to mortar sieved from the given concrete. The definition of
this setting time is arbitrary as there is no abrupt advent of setting in practice.1.73 The
Russians have attempted to define the setting time of concrete by the minimum
resistance between two embedded metal electrodes between which is passed a high-
frequency electric current.1.77
Soundness
It is essential that cement paste, once it has set, does not undergo a large change in
volume. In particular, there must be no appreciable expansion which, under conditions
of restraint, could result in a disruption of the hardened cement paste. Such expansion
may take place due to the delayed or slow hydration, or other reaction, of some
compounds present in the hardened cement, namely free lime, magnesia, and calcium
sulfate.
If the raw materials fed into the kiln contain more lime than can combine with the
acidic oxides, or if burning or cooling are unsatisfactory, the excess lime will remain
in a free condition. This hard-burnt lime hydrates only very slowly and, because

58
slaked lime occupies a larger volume than the original free calcium oxide, expansion
takes place. Cements which exhibit this expansion are described as unsound.
Lime added to cement does not produce unsoundness because it hydrates rapidly
before the paste has set. On the other hand, free lime present in clinker is
intercrystallized with other compounds and is only partially exposed to water during
the time before the paste has set.
Free lime cannot be determined by chemical analysis of cement because it is not
possible to distinguish between unreacted CaO and Ca(OH)2 produced by a partial
hydration of the calcium silicates when cement is exposed to the atmosphere. On the
other hand, a test on clinker, immediately after it has left the kiln, would show the free
lime content as no hydrated cement is then present.
A cement can also be unsound due to the presence of MgO, which reacts with
water in a manner similar to CaO. However, only periclase, that is, ‘dead-burnt’
crystalline MgO, is deleteriously reactive, and MgO present in glass is harmless. Up
to about 2 per cent of periclase (by mass of cement) combines with the main cement
compounds, but excess periclase generally causes expansion and can lead to slow
disruption.
Calcium sulfate is the third compound liable to cause expansion: in this case,
calcium sulfoaluminate is formed. It may be recalled that a hydrate of calcium sulfate
– gypsum – is added to cement clinker in order to prevent flash set, but if gypsum is
present in excess of the amount that can react with C3A during setting, unsoundness in
the form of a slow expansion will result. For this reason, standards limit very strictly
the amount of gypsum that can be added to clinker; the limits are well on the safe side
as far as the danger of unsoundness is concerned.1.46
Because unsoundness of cement is not apparent until after a period of months or
years, it is essential to test the soundness of cement in an accelerated manner: a test
devised by Le Chatelier is prescribed by BS EN 196-3 : 2005. The Le Chatelier
apparatus, shown in Fig. 1.26, consists of a small brass cylinder split along its
generatrix. Two indicators with pointed ends are attached to the cylinder on either
side of the split; in this manner, the widening of the split, caused by the expansion of
cement, is greatly magnified and can easily be measured. The cylinder is placed on a
glass plate, filled with cement paste of standard consistency, and covered with another
glass plate. The whole assembly is then placed in a cabinet at 20 ± 1 °C (68 ± 2 °F)
and a relative humidity of not less than 98 per cent. At the end of that period, the
distance between the indicators is measured, and the mould is immersed in water and
gradually brought to the boil in 30 minutes. After boiling for 3 hours, the assembly is
taken out and, after cooling, the distance between the indicators is again measured.
The increase in this distance represents the expansion of the cement, and for Portland
cements is limited to 10 mm by BS EN 197-1 : 2000. If the expansion exceeds this
value, a further test is made after the cement has been spread and aerated for 7 days.
During this time, some of the lime may hydrate or even carbonate, and a physical
breakdown in size may also take place. At the end of the 7-day period, the Le
Chatelier test is repeated, and the expansion of aerated cement must not exceed a
specified value, which in the past was 5 mm. A cement which fails to satisfy at least
one of these tests should not be used.

59
 Fig. 1.26. Le Chatelier apparatus
The Le Chatelier test detects unsoundness due to free lime only. Magnesia is rarely
present in large quantities in the raw materials from which cement is manufactured in
Great Britain, but it is encountered in other countries. An example is India, where
low-magnesia limestone occurs only to a limited extent. The bulk of cement there has,
therefore, a high MgO content but expansion can be significantly reduced by the
addition of active siliceous material such as fly ash or finely ground burnt clay.
Because of the importance of avoiding delayed expansion, in the United States, for
instance, soundness of cement is checked by the autoclave test, which is sensitive to
both free magnesia and free lime. In this test, prescribed by ASTM C 151-09, a neat
cement bar, 25 mm (or 1 in.) square in cross-section and with a 250 mm (or 10 in.)
gauge length, is cured in humid air for 24 hours. The bar is then placed in an
autoclave (a high-pressure steam boiler), which is raised to a temperature of 216 °C
(420 °F) (steam pressure of 2 ± 0.07 MPa (295 psi)) in 60 ± 15 min, and maintained at
this temperature for 3 hours. The high steam-pressure accelerates the hydration of
both magnesia and lime. The expansion of the bar due to autoclaving must not exceed
0.8 per cent.
The results of the autoclave test are affected not only by the compounds causing
expansion, but also by the C3A content, and by materials blended with the
cement,1.71 and are also subject to other anomalies. The test gives, therefore, no more
than a broad indication of the risk of long-term expansion in practice,1.1 but it is
generally overly severe as some MgO may remain inert; the test thus errs on the safe
side.1.86
No test is available for the detection of unsoundness due to an excess of calcium
sulfate, but its content can easily be determined by chemical analysis.
Strength of cement
The mechanical strength of hardened cement is the property of the material that is
perhaps most obviously required for structural use. It is not surprising, therefore, that
strength tests are prescribed by all specifications for cement.
The strength of mortar or concrete depends on the cohesion of the cement paste, on
its adhesion to the aggregate particles, and to a certain extent on the strength of the

60
aggregate itself. The last factor is not considered at this stage, and is eliminated in
tests on the quality of cement by the use of standard aggregates.
Strength tests are not made on a neat cement paste because of difficulties of
moulding and testing with a consequent large variability of test results. Cement–sand
mortar and, in some cases, concrete of prescribed proportions and made with specified
materials under strictly controlled conditions, are used for the purpose of determining
the strength of cement.
There are several forms of strength tests: direct tension, direct compression, and
flexure. The latter determines in reality the tensile strength in bending because, as is
well known, hydrated cement paste is considerably stronger in compression than in
tension.
In the past, the direct tension test on briquettes used to be commonly employed but
pure tension is rather difficult to apply so that the results of such a test show a fairly
large scatter. Moreover, since structural techniques are designed mainly to exploit the
good strength of concrete in compression, the direct tensile strength of cement is of
lesser interest than its compressive strength.
Similarly, flexural strength of concrete is generally of lesser interest than its
compressive strength, although in pavements the knowledge of the strength of
concrete in tension is of importance. In consequence, nowadays, it is the compressive
strength of cement that is considered to be crucial, and it is believed that the
appropriate test on cement is that on sand–cement mortar.
European Standard BS EN 196-1 : 2005 prescribes a compressive strength test on
mortar specimens. The specimens are tested as 40 mm equivalent cubes; they are
derived from 40 by 40 by 160 mm prisms, which are first tested in flexure so as to
break into halves, or are otherwise broken into halves. Thus an optional flexural
centre-point test over a span of 100 mm is possible.
The test is performed on mortar of fixed composition, made with a ‘CEN standard
sand’. (CEN is the acronym of the French name of the European Committee for
Standardization.) The sand is natural, siliceous, rounded sand which can be obtained
from various sources. Unlike Leighton Buzzard sand (see below), it is not of uniform
size but is graded between 80 μm and 1.6 mm. The sand/cement ratio is 3 and the
water/cement ratio is 0.50. The mortar is mixed in a cake mixer and compacted on a
jolting table with a drop of 15 mm; a vibrating table can also be used, provided it
results in similar compaction. The specimens are demoulded after 24 hours in a moist
atmosphere and thereafter cured in water at 20 °C.
Because the earlier British or similar tests are used in some countries, it may be
appropriate to give a brief description of those tests. Fundamentally, there are two
British standard methods of testing the compressive strength of cement: one uses
mortar, the other concrete.
In the mortar test, a 1 : 3 cement–sand mortar is used. The sand is standard
Leighton Buzzard sand, obtained from pits near a town of that name in Bedfordshire,
England; the sand is of single size. The mass of water in the mix is 10 per cent of the
mass of the dry materials. Expressed as a water/cement ratio, this corresponds to 0.40
by mass. A standard procedure, prescribed by BS 4550-3.4 : 1978, is followed in
mixing, and 70.7 mm (2.78 in.) cubes are made using a vibrating table with a
frequency of 200 Hz applied for two minutes. The cubes are demoulded after 24 hours
and further cured in water until tested in a wet-surface condition.

61
 The vibrated mortar test gives fairly reliable results but it has been suggested that
mortar made with single-size aggregate leads to a greater scatter of strength values
than would be obtained with concrete made under similar conditions. It can also be
argued that we are interested in the performance of cement in concrete and not in
mortar, especially one made with a single-size aggregate and never used in practice.
For these reasons, a test on concrete was introduced in British Standards.
In the concrete test, three water/cement ratios can be used, viz. 0.60, 0.55 and 0.45,
depending on the type of cement. The amounts of coarse and fine aggregate, which
have to come from particular quarries, are specified in BS 4550 : 4 and 5 : 1978.
Batches of 100 mm (or 4 in.) cubes are made by hand in a prescribed manner; the
temperature and humidity conditions of the mixing room, curing chamber,
compression testing room, and the temperature of the water curing tank are specified.
Apart from satisfying the minimum strength at specific ages, the strength at later ages
has to be higher than at an earlier age, because strength retrogression might be a sign
of unsoundness or other faults in the cement. The requirement of strength increase
with age applies also to the vibrated mortar cubes but it is not included in BS EN 197-
1 : 2000. The characteristic strength is specified. There are three classes: 32.5, 42.5,
and 52.5. These are values at 28 days, the latter two with a normal or early stength at
2 days. BS 1881 : 131 : 1998 specifies the test method on concrete cubes with made-
up aggregate whose grading is prescribed in BS EN 196-1 : 2005.
The ASTM method for testing the mean strength of cement is prescribed by ASTM
C 109-08 and uses a 1:2.75 mortar made with standard graded sand at a water/cement
ratio of 0.485; 50 mm (or 2 in.) cubes are tested.
It may be appropriate to consider the question: should tests on the strength of
cement be made on samples of cement paste, mortar or concrete? We have already
stated that specimens of neat cement paste are difficult to make. As far as concrete is
concerned, it is an appropriate medium for tests but the strength of concrete
specimens is influenced by the properties of the aggregate used. It would be difficult,
or even impractical, to use a standard aggregate for tests on concrete in various parts
of the country, let alone in different countries. The use of mortar with a reasonably
standard aggregate is a sensible compromise. In any case, all tests are comparative in
nature, rather than a direct measure of the compressive strength of hydrated cement
paste. Moreover, the influence of cement on the properties of mortar and concrete is
qualitatively the same, and the relation between the strengths of corresponding
specimens of the two materials is linear. This is shown, for instance, in Fig. 1.27:
mortar and concrete of fixed proportions, each with a water/cement ratio of 0.65, were
used. The strengths are not the same for the specimens of each pair, at least in part
because specimens of different shape and size were used, but there may also be an
inherent quantitative difference between the strengths of mortar and of concrete due to
the greater amount of entrapped air in mortar.

62
 Fig. 1.27. Relation between the strengths of concrete and of mortar of the same
water/cement ratio1.37
Another comparison of interest is that between the strength of concrete made to BS
4550-3.4 : 1978 with a water/cement ratio of 0.60 and the strength of mortar made to
BS EN 196-1 : 2005, with a water/cement ratio of 0.50. Not only the water/cement
ratio but also other conditions differ between the two tests, so that the resulting
strength values differ, too. Harrison1.88 found the following relation:
loge(M/C) = 0.28/d + 0.25
where C = compressive strength of BS 4550 concrete cubes in MPa
M = compressive strength of BS EN 196 mortar prisms in MPa
and d = age at test in days.
More conveniently, the ratio M/C can be tabulated as follows:

63
 In addition to the characteristics of the test specimens, there is an important
difference between the significance of the strength values obtained in the European
Standard BS EN 196-1 : 2005 and in the old British and most other standards: this
concerns the meaning of the term ‘minimum strength’. In the traditional standards, the
minimum value prescribed had to be exceeded by all test results. On the other hand, in
BS EN 196-1 : 2005, the minimum strength represents a characteristic value (see
p. 734) such that it is exceeded by 95 per cent of test results; in addition, there is laid
down an absolute value below which the specified strength must not fall.
References
1.1. F. M. LEA, The Chemistry of Cement and Concrete (London, Arnold, 1970).
1.2. R. H. BOGUE, Chemistry of Portland Cement (New York, Reinhold, 1955).
1.3. A. M. NEVILLE, Role of cement in creep of mortar, J. Amer. Concr. Inst., 55,
pp. 963–84 (March 1959).
1.4. M. A. SWAYZE, The quaternary system CaO–C5A3–C2F–C2S as modified by
saturation with magnesia, Amer. J. Sci., 244, pp. 65–94 (1946).
1.5. W. CZERNIN, Cement Chemistry and Physics for Civil Engineers (London,
Crosby Lockwood, 1962).
1.6. H. H. STEINOUR, The reactions and thermochemistry of cement hydration at
ordinary temperature, Proc. 3rd Int. Symp. on the Chemistry of Cement, pp.
261–89 (London, 1952).
1.7. R. H. BOGUE and W. LERCH, Hydration of portland cement
compounds, Industrial and Engineering Chemistry, 26, No. 8, pp. 837–47
(Easton, Pa., 1934).
1.8. E. P. FLINT and L. S. WELLS, Study of the system CaO–SiO2–H2O at 30 °C and
the reaction of water on the anhydrous calcium silicates, J. Res. Nat. Bur.
Stand., 12, No. 687, pp. 751–83 (1934).
1.9. S. GIERTZ-HEDSTROM, The physical structure of hydrated cements, Proc. 2nd
Int. Symp. on the Chemistry of Cement, pp. 505–34 (Stockholm, 1938).
1.10. T. C. POWERS, The non-evaporable water content of hardened portland cement
paste: its significance for concrete research and its method of
determination, ASTM Bul. No. 158, pp. 68–76 (May 1949).
1.11. L. S. BROWN and R. W. CARLSON, Petrographic studies of hydrated
cements, Proc. ASTM, 36, Part II, pp. 332–50 (1936).
1.12. L. E. COPELAND, Specific volume of evaporable water in hardened portland
cement pastes, J. Amer. Concr. Inst., 52, pp. 863–74 (1956).
1.13. S. BRUNAUER, J. C. HAYES and W. E. HASS, The heats of hydration of tricalcium
silicate and beta-dicalcium silicate, J. Phys. Chem., 58, pp. 279–87 (Ithaca,
NY, 1954).
1.14. F. M. LEA, Cement research: retrospect and prospect, Proc. 4th Int. Symp. on
the Chemistry of Cement, pp. 5–8 (Washington DC, 1960).
1.15. H. F. W. TAYLOR, Hydrated calcium silicates, Part I: Compound formation at
ordinary temperatures, J. Chem. Soc., pp. 3682–90 (London, 1950).
1.16. H. H. STEINOUR, The system CaO–SiO2–H2O and the hydration of the calcium
silicates, Chemical Reviews, 40, pp. 391–460 (USA, 1947).

64
 1.17. J. W. T. SPINKS, H. W. BALDWIN and T. THORVALDSON, Tracer studies of
diffusion in set Portland cement, Can. J. Technol., 30, Nos 2 and 3, pp. 20–8
(1952).
1.18. M. NAKADA, Process operation and environmental protection at the Yokoze
cement works, Zement-Kalk-Gips, 29, No. 3, pp. 135–9 (1976).
1.19. S. DIAMOND, C/S mole ratio of C-S-H gel in a mature C3S paste as determined
by EDXA, Cement and Concrete Research, 6, No. 3, pp. 413–16 (1976).
1.20. J. D. BERNAL, J. W. JEFFERY and H. F. W. TAYLOR, Crystallographic research on
the hydration of Portland cement: A first report on investigations in
progress, Mag. Concr. Res., 3, No. 11, pp. 49–54 (1952).
1.21. W. C. HANSEN, Discussion on “Aeration cause of false set in portland
cement”, Proc. ASTM, 58, pp. 1053–4 (1958).
1.22. T. C. POWERS, The physical structure and engineering properties of
concrete, Portl. Cem. Assoc. Res. Dept. Bul. 39 pp. (Chicago, July 1958).
1.23. T. C. POWERS, A discussion of cement hydration in relation to the curing of
concrete, Proc. Highw. Res. Bd., 27, pp. 178–88 (Washington, 1947).
1.24. T. C. POWERS and T. L. BROWNYARD, Studies of the physical properties of
hardened Portland cement paste (Nine parts), J. Amer. Concr. Inst., 43 (Oct.
1946 to April 1947).
1.25. G. J. VERBECK, Hardened concrete – pore structure, ASTM Sp. Tech. Publ. No.
169, pp. 136–42 (1955).
1.26. T. C. POWERS, L. E. COPELAND and H. M. MANN, Capillary continuity or
discontinuity in cement pastes, J. Portl. Cem. Assoc. Research and
Development Laboratories, 1, No. 2, pp. 38–48 (May 1959).
1.27. T. C. POWERS, Structure and physical properties of hardened portland cement
paste, J. Amer. Ceramic Soc., 41, pp. 1–6 (Jan. 1958).
1.28. L. E. COPELAND and J. C. HAYES, Porosity of hardened portland cement
pastes, J. Amer. Concr. Inst., 52, pp. 633–40 (Feb. 1956).
1.29. R. W. CARLSON and L. R. FORBRICK, Correlation of methods for measuring heat
of hydration of cement, Industrial and Engineering Chemistry (Analytical
Edition), 10, pp. 382–6 (Easton, Pa., 1938).
1.30. W. LERCH and C. L. FORD, Long-time study of cement performance in
concrete, Chapter 3: Chemical and physical tests of the cements, J. Amer.
Concr. Inst., 44, pp. 743–95 (April 1948).
1.31. N. DAVEY and E. N. Fox, Influence of temperature on the strength
development of concrete, Build. Res. Sta. Tech. Paper No. 15 (London,
HMSO, 1933).
1.32. W. LERCH and R. H. BOGUE, Heat of hydration of portland cement pastes, J.
Res. Nat. Bur. Stand., 12, No. 5, pp. 645–64 (May 1934).
1.33. H. WOODS, H. H. STEINOUR and H. R. STARKE, Heat evolved by cement in
relation to strength, Engng News Rec., 110, pp. 431–3 (New York, 1933).
1.34. G. J. VERBECK and C. W. FOSTER, Long-time study of cement performance in
concrete, Chapter 6: The heats of hydration of the cements, Proc. ASTM, 50,
pp. 1235–57 (1950).

65
 1.35. H. WOODS, H. R. STARKE and H. H. STEINOUR, Effect of cement composition on
mortar strength, Engng News Rec., 109, No. 15, pp. 435–7 (New York,
1932).
1.36. R. E. DAVIS, R. W. CARLSON, G. E. TROXELL and J. W. KELLY, Cement
investigations for the Hoover Dam, J. Amer. Concr. Inst., 29, pp. 413–31
(1933).
1.37. S. WALKER and D. L. BLOEM, Variations in portland cement, Proc. ASTM, 58,
pp. 1009–32 (1958).
1.38. F. M. LEA, The relation between the composition and properties of Portland
cement, J. Soc. Chem. Ind., 54, pp. 522–7 (London, 1935).
1.39. F. M. LEA and F. E. JONES, The rate of hydration of Portland cement and its
relation to the rate of development of strength, J. Soc. Chem. Ind., 54, No.
10, pp. 63–70T (London, 1935).
1.40. L. S. BROWN, Long-time study of cement performance in concrete, Chapter 4:
Microscopical study of clinkers, J. Amer. Concr. Inst., 44, pp. 877–923
(May 1948).
1.41. W. LERCH, Approximate glass content of commercial Portland cement
clinker, J. Res. Nat. Bur. Stand., 20, pp. 77–81 (Jan. 1938).
1.42. F. M. LEA, Cement and Concrete, Lecture delivered before the Royal
Institute of Chemistry, London, 19 Dec. 1944 (Cambridge, W. Heffer and
Sons, 1944).
1.43. W. H. PRICE, Factors influencing concrete strength, J. Amer. Concr. Inst., 47,
pp. 417–32 (Feb. 1951).
1.44. US BUREAU OF RECLAMATION, Investigation into the effects of cement fineness
and alkali content on various properties of concrete and mortar, Concrete
Laboratory Report No. C-814 (Denver, Colorado, 1956).
1.45. S. BRUNAUER, P. H. EMMETT and E. TELLER, Adsorption of gases in multi-
molecular layers, J. Amer. Chem. Soc., 60, pp. 309–19 (1938).
1.46. W. LERCH, The influence of gypsum on the hydration and properties of
portland cement pastes, Proc. ASTM, 46, pp. 1252–92 (1946).
1.47. L. E. COPELAND and R. H. BRAGG, Determination of Ca(OH)2 in hardened
pastes with the X-ray spectrometer, Portl. Cem. Assoc. Rep. (Chicago, 14
May 1953).
1.48. L. E. COPELAND and J. C. HAYES, The determination of non-evaporable water
in hardened portland cement paste, ASTM Bul. No. 194, pp. 70–4 (Dec.
1953).
1.49. L. E. COPELAND, D. L. KANTRO and G. VERBECK, Chemistry of hydration of
portland cement, Proc. 4th Int. Symp. on the Chemistry of Cement, pp. 429–
65 (Washington DC, 1960).
1.50. P. SELIGMANN and N. R. GREENING, Studies of early hydration reactions of
portland cement by X-ray diffraction, Highway Research Record, No. 62, pp.
80–105 (Highway Research Board, Washington DC, 1964).
1.51. W. G. HIME and E. G. LABONDE, Particle size distribution of portland cement
from Wagner turbidimeter data, J. Portl. Cem. Assoc. Research and
Development Laboratories, 7, No. 2, pp. 66–75 (May 1965).

66
 1.52. C. D. LAWRENCE, The properties of cement paste compacted under high
pressure, Cement Concr. Assoc. Res. Rep. No. 19 (London, June 1969).
1.53. R. F. FELDMAN and P. J. SEREDA, A model for hydrated Portland cement paste
as deduced from sorption-length change and mechanical
properties, Materials and Structures, No. 6, pp. 509–19 (Nov.–Dec. 1968).
1.54. P. SELIGMANN, Nuclear magnetic resonance studies of the water in hardened
cement paste, J. Portl. Cem. Assoc. Research and Development
Laboratories, 10, No. 1, pp. 52–65 (Jan. 1968).
1.55. G. VERBECK, Cement hydration reactions at early ages, J. Portl. Cem. Assoc.
Research and Development Laboratories, 7, No. 3, pp. 57–63 (Sept. 1965).
1.56. R. L. BLAINE, H. T. ARNI and M. R. DEFORE, Interrelations between cement
and concrete properties, Part 3, Nat. Bur. Stand. Bldg Sc. Series
8 (Washington DC, April 1968).
1.57. M. VON EUW and P. GOURDIN, Le calcul prévisionnel des résistances des
ciments Portland, Materials and Structures, 3, No. 17, pp. 299–311 (Sept.–
Oct. 1970).
1.58. W. J. MCCOY and D. L. ESHENOUR, Significance of total and water soluble
alkali contents of cement, Proc. 5th Int. Symp. on the Chemistry of
Cement, 2, pp. 437–43 (Tokyo, 1968).
1.59. M. DOHNALIK and K. FLAGA, Nowe spostrzezenia w problemie czasu wiazania
cementu, Archiwum Inzynierii Ladowej, 16, No. 4, pp. 745–52 (1970).
1.60. S. DIAMOND, Cement paste microstructure – an overview at several
levels, Proc. Conf. Hydraulic Cement Pastes: Their Structure and
Properties, pp. 2–30 (Sheffield, Cement and Concrete Assoc., April 1976).
1.61. J. F. YOUNG, A review of the mechanisms of set-retardation in Portland
cement pastes containing organic admixtures. Cement and Concrete
Research, 2, No. 4, pp. 415–33 (July 1972).
1.62. S. BRUNAUER, J. SKALNY, I. ODLER and M. YUDENFREUND, Hardened portland
cement pastes of low porosity. VII. Further remarks about early hydration.
Composition and surface area of tobermorite gel, Summary, Cement and
Concrete Research, 3, No. 3, pp. 279–94 (May 1973).
1.63. D. WINSLOW and S. DIAMOND, Specific surface of hardened portland cement
paste as determined by small angle X-ray scattering, J. Amer. Ceramic
Soc., 57, pp. 193–7 (May 1974).
1.64. F. H. WITTMANN and F. SCHLUDE, Microwave absorption of hardened cement
paste, Cement and Concrete Research, 5, No. 1, pp. 63–71 (Jan. 1975).
1.65. I. ODLER, M. YUDENFREUND, J. SKALNY and S. BRUNAUER, Hardened Portland
cement pastes of low porosity. III. Degree of hydration. Expansion of paste.
Total porosity, Cement and Concrete Research, 2, No. 4, pp. 463–81 (July
1972).
1.66. V. S. RAMACHANDRAN and C.-M. ZHANG, Influence of CaCO3, Il Cemento, 3, pp.
129–52 (1986).
1.67. D. KNÖFEL, Quantitative röntgenographische Freikalkbestimmung zur
Produktions-kontrolle im Zementwerk, Zement-Kalk-Gips, 23, No. 8, pp.
378–9 (Aug. 1970).

67
 1.68. T. KNUDSEN, Quantitative analysis of the compound composition of cement
and cement clinker by X-ray diffraction, Amer. Ceramic Soc. Bul., 55, No.
12, pp. 1052–5 (Dec. 1976).
1.69. J. OLEK, M. D. COHEN and C. LOBO, Determination of surface area of portland
cement and silica fume by mercury intrusion porosimetry, ACI Materials
Journal, 87, No. 5, pp. 473–8 (1990).
1.70. S. M. KHALIL and M. A. WARD, Influence of a lignin-based admixture on the
hydration of Portland cements, Cement and Concrete Research, 3, No. 6, pp.
677–88 (Nov. 1973).
1.71. J. CALLEJA, L’expansion des ciments, Il Cemento, 75, No. 3, pp. 153–64
(July–Sept. 1978).
1.72. J. F. YOUNG et al., Mathematical modelling of hydration of cement: hydration
of dicalcium silicate, Materials and Structures, 20, No. 119, pp. 377–82
(1987).
1.73. J. H. SPROUSE and R. B. PEPPLER, Setting time, ASTM Sp. Tech. Publ. No. 169B,
pp. 105–21 (1978).
1.74. I. ODLER and H. DÖRR, Early hydration of tricalcium silicate. I. Kinetics of the
hydration process and the stoichiometry of the hydration products, Cement
and Concrete Research, 9, No. 2, pp. 239–48 (March 1979).
1.75. M. H. WILLS, Accelerated strength tests, ASTM Sp. Tech. Publ. No. 169B, pp.
162–79 (1978).
1.76. J. BROTSCHI and P. K. MEHTA, Test methods for determining potential alkali–
silica reactivity in cements, Cement and Concrete Research, 8, No. 2, pp.
191–9 (March 1978).
1.77. RILEM NATIONAL COMMITTEE OF THE USSR, Method of determination of the
beginning of concrete setting time, U.S.S.R. Proposal to RILEM Committee
CPC-14, 7 pp. (Moscow, July 1979).
1.78. R. SERSALE, R. CIOFFI, G. FRIGIONE and F. ZENONE, Relationship between
gypsum content, porosity and strength in cement, Cement and Concrete
Research, 21, No. 1, pp. 120–6 (1991).
1.79. I. ODLER, Strength of cement (Final Report), Materials and Structures, 24,
No. 140, pp. 143–57 (1991).
1.80. ANON, Saving money in cement production, Concrete International, 10, No.
1, pp. 48–9 (1988).
1.81. G. C. BYE, Portland Cement: Composition, Production and Properties, 149
pp. (Oxford, Pergamon Press, 1983).
1.82. Z. BERHANE, Heat of hydration of cement pastes, Cement and Concrete
Research, 13, No. 1, pp. 114–18 (1983).
1.83. M. KAMINSKI and W. ZIELENKIEWICZ, The heats of hydration of cement
constituents, Cement and Concrete Research, 12, No. 5, pp. 549–58 (1982).
1.84. H. F. W. TAYLOR, Modification of the Bogue calculation, Advances in Cement
Research, 2, No. 6, pp. 73–9 (1989).
1.85. F. P. GLASSER, Progress in the immobilization of radioactive wastes in
cement, Cement and Concrete Research, 22, Nos 2/3, pp. 201–16 (1992).

68
 1.86. V. S. RAMACHANDRAN, A test for “unsoundness” of cements containing
magnesium oxide, Proc. 3rd Int. Conf. on the Durability of Building
Materials and Components, Espoo, Finland, 3, pp. 46–54 (1984).
1.87. J. J. BEAUDOIN and V. S. RAMACHANDRAN, A new perspective on the hydration
characteristics of cement phases, Cement and Concrete Research, 22, No. 4,
pp. 689–94 (1992).
1.88. T. A. HARRISON, New test method for cement strength, BCA Eurocements,
Information Sheet No. 2, 2 pp. (Nov. 1992).
1.89. D. M. ROY and G. R. GOUDA, Optimization of strength in cement
pastes, Cement and Concrete Research, 5, No. 2, pp. 153–62 (1975).
1.90. F. MASSAZZA and M. TESTOLIN, Latest developments in the use of admixtures
for cement and concrete, Il Cemento, 77, No. 2, pp. 73–146 (1980).
1.91. P.-C. AÏTCIN, S. L. SARKAR, M. REGOURD and D. VOLANT, Retardation effect of
superplasticizer on different cement fractions. Cement and Concrete
Research, 17, No. 6, pp. 995–9 (1987).
1.92. A. NONAT and J. C. MUTIN (eds), Hydration and setting of cements, Proc. of
Int. RILEM Workshop on Hydration, Université de Dijon, France, 418 pp.
(London, Spon, 1991).
1.93. M. RELIS, W. B. LEDBETTER and P. HARRIS, Prediction of mortar-cube strength
from cement characteristics, Cement and Concrete Research, 18, No. 5, pp.
674–86 (1988).
1.94. J. BARON and R. SANTERAY (Eds), Le Béton Hydraulique – Connaissance et
Pratique, 560 pp. (Presses de l’Ecole Nationale des Ponts et Chaussées,
Paris, 1982).
1.95. B. OSBÆCK, On the influence of alkalis on strength development of blended
cements, in The Chemistry and Chemically Related Properties of Cement,
British Ceramic Proceedings, No. 35, pp. 375–83 (Sept. 1984).
1.96. H. BRAUN, Produktion, Energieeinsatz und Emissionen im Bereich der
Zementindustrie, Zement + Beton, pp. 32–34 (Jan. 1994).
1.97. D. P. BENTZ et al., Limestone fillers conserve cement, ACI Journal, 31, No.
11, pp. 41–6 (2009).

69
Chapter 2. Cementitious materials of different types

The previous chapter dealt with the properties of Portland cement in general, and we
have seen that cements differing in chemical composition and physical characteristics
may exhibit different properties when hydrated. It should thus be possible to select
mixtures of raw materials for the production of cements with various desired
properties. In fact, several types of Portland cement are available commercially and
additional special cements can be produced for specific uses. Several non-Portland
cements are also available.
Before describing the various types of Portland cement, a more general discussion
of the cementitious materials used in concrete may be useful.

Categorization of cementitious materials*

*
This section was substantially published in ref. 2.5.
Originally, concrete was made using a mixture of only three materials: cement,
aggregate, and water; almost invariably, the cement was Portland cement, as
discussed in Chapter 1. Later on, in order to improve some of the properties of
concrete, either in the fresh or in the hardened state, very small quantities of chemical
products were added into the mix. These chemical admixtures, often called simply
admixtures, are discussed in Chapter 4.
Later still, other materials, inorganic in nature, were introduced into the concrete
mix. The original reasons for using these materials were usually economic: they were
cheaper than Portland cement, sometimes because they existed as natural deposits
requiring no, or little, processing, sometimes because they were a byproduct or waste
from industrial processes. A further spur to the incorporation of these ‘supplementary’
materials in the concrete mix was given by the sharp increase in the cost of energy in
the 1970s, and we recall that the cost of energy represents a major proportion of the
cost of the production of cement (see p. 7).
Yet further encouragement for the use of some of the ‘supplementary’ materials
was provided by the ecological concerns about opening of pits and quarries for the
raw materials required for the production of Portland cement on the one hand and, on
the other, about the means of disposal of the industrial waste materials such as
blastfurnace slag, fly ash, or silica fume. Moreover, the manufacture of Portland
cement itself is ecologically harmful in that the production of one tonne of cement
results in about one tonne of carbon dioxide being discharged into the atmosphere.
It would be incorrect to infer from the previous, historical account that the
supplementary materials were introduced into concrete solely by the ‘push’ of their
availability. These materials also bestow various desirable properties on concrete,
sometimes in the fresh state, but more often in the hardened state. This ‘pull’,
combined with the ‘push’, has resulted in a situation such that, in very many countries,
a high proportion of concrete contains one or more of these supplementary materials.
It is therefore inappropriate to consider them, as was sometimes done in the past, as
cement replacement materials or as ‘extenders’.
If, as just stated, the materials which we have hitherto described as supplementary
are, in their own right, proper components of the cementitious materials used in
making concrete, then a new terminology has to be sought. No single terminology has

70
been agreed or accepted on a world-wide basis, and it may be useful briefly to discuss
the nomenclature used in various publications.
In so far as concrete is concerned, the cementitious material always contains
Portland cement of the traditional variety, that is ‘pure’ Portland cement. Therefore,
when other materials are also included, it is possible to refer to the ensemble of the
cementitious materials used as Portland composite cements. This is a logical term,
and so is the term blended Portland cements.
The European approach of BS EN 197-1 : 2000 is to use the term CEM cement,
which requires the presence of the Portland cement component; thus CEM cement by
implication excludes high-alumina cement. The name CEM cement is not thought to
be explicit or of general appeal. There are 27 common cements in five categories,
CEM I to V.
The current American approach is given in ASTM C 1157-10, which covers
blended hydraulic cements for both general and special applications. A blended
hydraulic cement is defined as follows: “A hydraulic cement consisting of two or
more inorganic constituents which contribute to the strength-gaining properties of the
cement, with or without other constituents, processing additions and functional
additions.”
This terminology is sound except that the term ‘inorganic constituent’ is difficult
to relate to the actual materials incorporated in concrete, typically natural or
industrially produced pozzolana, fly ash, silica fume, or ground granulated
blastfurnace slag. Moreover, emphasis on the term ‘hydraulic’ may conjure up a
wrong image in the eyes of the general users of cement. Furthermore, the ASTM
terminology is not used by the American Concrete Institute.
The preceding, rather lengthy, discussion explains the difficulty of classifying and
categorizing the different materials involved. The situation is not helped by a lack of
international nomenclature. Indeed, more than one approach is possible but the
difficulty is exacerbated by the fact that some of the divisions are not mutually
exclusive.
In view of the international use of this book, it has been decided to use the
following terminology.
A cement consisting of Portland cement with no more than 5 per cent of another
inorganic material will be referred to as Portland cement. We should recall that prior
to 1991, Portland cements were generally expected to be ‘pure’, that is, not to contain
minor additions other than gypsum or grinding aids.
A cement consisting of Portland cement and one or more appropriate inorganic
materials will be called blended cement. This term is close to that used in ASTM C
1157-10. Like ASTM, we use the term ‘blended’ to include both the results of
blending the separate powders and of intergrinding the parent materials, e.g. Portland
cement clinker and ground granulated blastfurnace slag (see p. 79).
There is some difficulty in choosing the term for the components which make up a
blended cement. The terms ‘constituent’ and ‘component’ run the risk of confusion
with the chemical compounds in Portland cement. What all the materials with which
we are concerned have in common is that, in the words of ASTM C 1157-10, they
“contribute to the strength-gaining properties of the cement”. In actual fact, some of
these materials are cementitious in themselves, some have latent cementitious
properties, yet others contribute to the strength of concrete primarily through their

71
physical behaviour. It is proposed, therefore, to refer to all these materials
as cementitious materials. Purists might criticize this choice, but it has the important
merits of simplicity and clarity.
The individual cementitious materials will be discussed later in this chapter but, for
convenience, Table 2.1 describes their relevant properties; it can be seen that there are
no clear-cut divisions with respect to hydraulic, that is, truly cementitious, properties.

Table 2.1. Cementitious Nature of Materials in Blended Cements

As already mentioned, all the cementitious materials, as just defined, have one
property in common: they are at least as fine as the particles of Portland cement, and
sometimes much finer. Their other features, however, are diverse. This applies to their
origin, their chemical composition, and their physical characteristics such as surface
texture or specific gravity.
There are several ways of preparing a blended cement. One way is to integrind the
other cementitious materials with the cement clinker so that an integral blended
cement is produced. The second way is for two or, more rarely, three materials in their
final form to be truly blended. Alternatively, Portland cement and one or more
cementitious materials can be separately, but simultaneously or nearly so, fed into the
concrete mixer.
Furthermore, the relative amounts of Portland cement and of the other
cementitious materials in the concrete mix vary widely: sometimes the proportion of
the other cementitious materials is low, in other mixes they constitute a significant
proportion, even a major part, of the blended cement.
Thus, in this book, the term ‘cementitious material’ will be used for all the powder
material, other than that which forms the finest particles of aggregate, provided that
one of the powder materials is cement. With very few exceptions considered on
p. 82 and 91, the cement is Portland cement. Thus, the cementitious material may be
Portland cement alone or it may comprise Portland cement and one or more other
cementitious materials.
A given cementitious material may be hydraulic in nature, that is, it may undergo
hydration on its own and contribute to the strength of the concrete. Alternatively, it
may have latent hydraulic properties: that is, it may exhibit hydraulic activity only in
consequence of chemical reaction with some other compounds such as the products of

72
hydration of Portland cement which co-exists in the mixture. Yet a third possibility is
for the cementitious material to be largely chemically inert but to have a catalytic
effect on the hydration of other materials, e.g. by fostering nucleation and densifying
the cement paste, or to have a physical effect on the properties of the fresh concrete.
Materials in this category are called fillers. Fillers will be discussed on p. 88.
For the benefit of American readers, it should be mentioned that the term “mineral
admixtures”, used by the American Concrete Institute to describe the non-hydraulic
supplementary materials, will not be used in this book. The word “admixture”
conjures up a minor component, something added to the ‘main mix’, and yet, as
already mentioned, some of the ‘supplementary’ materials are present in large
proportions.
The different categories of cementitious materials will be discussed later in this
chapter. Their more specific uses and their detailed influence on the properties of
concrete will be considered, as appropriate, throughout the book.
Different cements
In the preceding section, we discussed cementitious materials on the basis of their
broad composition and rational classification. For practical purposes of selection of an
appropriate Portland cement or a blended cement, it is useful to consider a
classification based on the relevant physical or chemical property, such as a rapid gain
of strength, low rate of evolution of the heat of hydration, or resistance to sulfate
attack.
In order to facilitate the discussion, a list of different Portland cements, with or
without other cementitious materials, together with the American description
according to ASTM Standards C 150-09 or C 595-10, where available, is given
in Table 2.2. The former ASTM composition limits for some of these cements have
already been listed (Table 1.9), and typical, historical values of compound
composition are given in Table 2.3.2.34

Table 2.2. Main Types of Portland Cement

73
 Table 2.3. Typical Values of Compound Composition of Portland Cements of
Different Types2.34

74
 The unification of standards within the European Union, including also some other
European countries, has led to the first common standard for cement published by the
European Committee for Standardization, namely, BS EN 197-1 : 2000 “Cement–
composition, specifications and conformity criteria for common cements”. A
simplified version of the classification used in that standard is given in Table 2.4.

Table 2.4. Classification of main cements according to European Standard BS

EN 197-1 : 2000

Many of the cements have been developed to ensure good durability of concrete
under a variety of conditions. It has not been possible, however, to find in the
composition of cement a complete answer to the problem of durability of concrete:
the principal mechanical properties of hardened concrete, such as strength, shrinkage,
permeability, resistance to weathering, and creep, are affected also by factors other
than cement composition, although this determines to a large degree the rate of gain
of strength.2.2 Figure 2.1 shows the rate of development of strength of concretes made
with cements of different types: while the rates vary considerably, there is little

75
difference in the 90-day strength of cements of all types;2.1 in some cases, e.g. Fig. 2.2,
the differences are greater.2.4 The general tendency is for the cements with a low rate
of hardening to have a slightly higher ultimate strength. For instance, Fig. 2.1 shows
that Type IV cement has the lowest strength at 28 days but develops the second
highest strength at the age of 5 years. A comparison of Figs 2.1 and 2.2 illustrates the
fact that differences between cement types are not readily quantified.

Fig. 2.1. Strength development of concretes containing 335 kg of cement per

cubic metre (565 lb/yd3) and made with cements of different types2.1

76
 Fig. 2.2. Strength development of concretes with a water/cement ratio of 0.49
made with cements of different types2.4
Still referring to Fig. 2.2, we should note that the retrogression of strength of
concrete made with Type II cement is not characteristic of this type of cement. The
pattern of low early and high late strength agrees with the influence of the ultitial
framework of hardened cement on the ultimate development of strength: the more
slowly the framework is established the denser the gel and the higher the ultimate
strength. Nevertheless, significant differences in the important physical properties of
cements of different types are found only in the earlier stages of hydration:2.3 in well-
hydrated pastes the differences are only minor.
The division of cements into different types is necessarily no more than a broad
functional classification, and there may sometimes be wide differences between
cements of nominally the same type. On the other hand, there are often no sharp
discontinuities in the properties of different types of cement, and many cements can
be classified as more than one type.
Obtaining some special property of cement may lead to undesirable features in
another respect. For this reason, a balance of requirements may be necessary, and the
economic aspect of manufacture must also be considered. Type II cement is an
example of a ‘compromise’ all-round cement.
The methods of manufacture have improved steadily over the years, and there has
been a continual development of cements to serve different purposes with a
corresponding change in specifications. On the other hand, some of the changes
proved to be disadvantageous when they were not accompanied by a change in
concrete practice; this is discussed on p. 335.

77
Ordinary Portland cement
This is by far the most common cement in use: about 90 per cent of all cement used in
the United States (total production in 2008 of about 73 million tonnes per annum) and
a like percentage of the ordinary type in the United Kingdom (total production of 12
million tonnes per annum in 2005). It may be interesting to note that in 2007 the
annual consumption of cement in the United Kingdom was equivalent to nearly 250
kg per head of population: the corresponding figure for the United States was 360 kg.
For every man, woman and child in the world, the consumption in 2007 was 420 kg
per annum, which is second only to the consumption of water. The biggest change
occurred in China, where the increase between 1995 and 2004 was 90%, and the
current consumption represents over 50% of the world production. The global
production is forecast to reach 3.5 billion tonnes in 2013. With the considerable
increase in the use of fly ash as a cementitious material, the quantity of concrete used
is no longer proportional to consumption of Portland cement.
Ordinary Portland (Type I) cement is admirably suitable for use in general
concrete construction when there is no exposure to sulfates in the soil or groundwater.
The specification for this cement is given in European Standard BS EN 197-1 : 2000.
In keeping with the modern trend towards performance-oriented specifications, little
is laid down about the chemical composition of the cement, either in terms of
compounds or of oxides. Indeed, the standard requires only that it is made from 95 to
100 per cent of Portland cement clinker and 0 to 5 per cent of minor additional
constituents, all by mass, the percentages being those of the total mass except calcium
sulfate and manufacturing additives such as grinding aids.
The limitation on the clinker composition is that not less than two-thirds of its
mass consists of C3S and C2S taken together, and that the ratio of CaO to SiO2, also by
mass, be not less than 2.0. The content of MgO is limited to a maximum of 5.0 per
cent.
The minor additional constituents, referred to above, are one or more of the other
cementitious materials (see p. 64) or a filler. A filler is defined as any natural or
inorganic material other than a cementitious material. An example of a filler is a
calcareous material which, due to its particle distribution, improves the physical
properties of the cement, for example, workability or water retention. Fillers are
discussed more fully on p. 88.
Thus, BS EN 197-1 : 2000 contains no detailed requirements about the proportion
of the various oxides in the clinker which were included in the previous versions of
British Standards. As some of those requirements are still used in many countries, it is
useful to mention the lime saturation factor which is to be not greater than 1.02 and
not less than 0.66. For cement, the factor is defined as:

where each term in brackets denotes the percentage by mass of the given compound
present in the cement.
The upper limit of the lime saturation factor ensures that the amount of lime is not
so high as to result in free lime appearing at the clinkering temperature in equilibrium
with the liquid present. The unsoundness of cement caused by free lime was discussed
in the previous chapter, and is indeed controlled by the Le Chatelier test. Too low a

78
lime saturation factor would make the burning in the kiln difficult and the proportion
of C3S in the clinker would be too low for the development of early strength.
Methods of chemical analysis of cement are prescribed in European Standard BS
EN 196-2 : 2005.
As British Standard BS 12 : 1996 (withdrawn 2000) is still in use in some
countries, it should be mentioned that it limits the expansion in the Le Chatelier test,
determined in accordance with BS EN 196-3 : 2005 to not more than 10 mm. Further
requirements of BS 12 : 1996 and BS EN 197-1 : 2000, which replaced BS 12 : 1996,
are: the SO3 content of not more than 3.5 or 4.0 per cent; and chloride content of not
more than 0.10 per cent. Limits on the insoluble residue and the loss on ignition are
also given. ASTM C 150-09 does not specify any limits on the SO3 content.
British Standard BS 12 : 1996 classifies Portland cements according to their
compressive strength, as shown in Table 2.5. The 28-day minimum strength in MPa
gives the name of the class: 32.5, 42.5, 52.5, and 62.5. The 28-day strengths of the
two lower classes are prescribed by a range, that is, each class of cement has a
maximum value of strength as well as a minimum. Moreover, cements of class 32.5
and 42.5 are each subdivided into two subclasses, one with an ordinary early strength,
the other with a high early strength. The two subclasses with a high early strength,
denoted by the letter R, are rapid-hardening cements, and they will be considered in
the next section.

Table 2.5. Compressive Strength Requirements of Cement According to BS 12 :

1996

The advantage of prescribing the class 32.5 and 42.5 cements by a range of
strength of 20 MPa is that, during construction, wide variations in strength, especially
downwards, are avoided. Furthermore, and perhaps more importantly, an excessively
high strength at the age of 28 days would allow, as was the case in the 1970s and
1980s, a specified strength of concrete to be achieved at an unduly low cement
content. This topic is considered more fully on p. 335.
Rapid-hardening Portland cement
This cement comprises Portland cement subclasses of 32.5 and 42.5 MPa as
prescribed by BS EN 197-1 : 2000. Rapid-hardening Portland cement (Type III), as its

79
name implies, develops strength more rapidly, and should, therefore, be correctly
described as high early strength cement. The rate of hardening must not be confused
with the rate of setting: in fact, ordinary and rapid-hardening cements have similar
setting times, prescribed by BS 12 : 1996 as an initial setting time of not less than 45
minutes. The final setting time is no longer prescribed. BS EN 197-1 : 2000 does not
prescribe fineness.
The increased rate of gain of strength of the rapid-hardening Portland cement is
achieved by a higher C3S content (higher than 55 per cent, but sometimes as high as
70 per cent) and by a finer grinding of the cement clinker. British Standard BS 12 :
1996, unlike previous versions of BS 12, does not prescribe the fineness of cement,
either ordinary or rapid-hardening. However, the standard provides for an
optional controlled fineness Portland cement and so does BS EN 197-1 : 2000. The
range of fineness is agreed between the manufacturer and the user. Such cement is
valuable in applications where it makes it easier to remove excess water from the
concrete during compaction because the fineness is more critical than the compressive
strength.
In practice, rapid-hardening Portland cement has a higher fineness than ordinary
Portland cement. Typically, ASTM Type III cements have a specific surface,
measured by the Blaine method, of 450 to 600 m2/kg, compared with 300 to 400 m2/kg
for Type I cement. The higher fineness significantly increases the strength at 10 to 20
hours, the increase persisting up to about 28 days. Under wet curing conditions, the
strengths equalize at the age of 2 to 3 months, but later on the strength of the cements
with a lower fineness surpasses that of the high fineness cements.2.9
This behaviour should not be extrapolated to cements with a very high fineness,
which increase the water demand of the mix. In consequence, at a given cement
content and for a given workability, the water/cement ratio is increased and this
offsets the benefits of the higher fineness with respect to early strength.
The requirements of soundness and chemical properties are the same for rapid-
hardening as for ordinary Portland cement and need not, therefore, be repeated.
The use of rapid-hardening cement is indicated where a rapid strength
development is desired, e.g. when formwork is to be removed early for re-use, or
where sufficient strength for further construction is wanted as quickly as practicable.
Rapid-hardening cement is not much dearer than ordinary cement but it accounts for
only a few per cent of all cement manufactured in the United Kingdom and in the
United States. Because, however, the rapid gain of strength means a high rate of heat
development, rapid-hardening Portland cement should not be used in mass
construction or in large structural sections. On the other hand, for construction at low
temperatures the use of cement with a high rate of heat evolution may prove a
satisfactory safeguard against early frost damage.
Special very rapid-hardening Portland cements
There exist several specially manufactured cements which are particularly rapid
hardening. One of these, a so-called ultra high early strength cement. This type of
cement is not standardized but rather supplied by individual cement manufacturers.
Generally, the rapid strength development is achieved by grinding the cement to a
very high fineness: 700 to 900 m2/kg. Because of this, the gypsum content has to be
higher (4 per cent expressed as SO3) than in cements complying with BS EN 197-1 :
2000, but in all other respects the ultra high early strength cement satisfies the

80
requirements of that standard. It can be noted that the high gypsum content has no
adverse effect on long-term soundness as the gypsum is used up in the early reactions
of hydration.
The effect of the fineness of cement on the development of strength is illustrated
in Fig. 2.3. All the cements used in this study2.19 had the C3S content between 45 and
48 per cent, and the C3A content between 14.3 and 14.9 per cent.

Fig. 2.3. Increase in strength of concrete with a water/cement ratio of 0.40 using
Portland cements of varying specific surface (determined by the air permeability
method)2.19
Ultra high early strength cement is manufactured by separating fines from rapid-
hardening Portland cement by a cyclone air elutriator. Because of its high fineness,
the ultra high early strength cement has a low bulk density and deteriorates rapidly on
exposure. High fineness leads to rapid hydration, and therefore to a high rate of heat
generation at early ages and to a rapid strength development; for instance, the 3-day
strength of rapid-hardening Portland cement is reached at 16 hours, and the 7-day
strength at 24 hours.2.35 There is, however, little gain in strength beyond 28 days.
Typical strengths of 1:3 concretes made with the ultra high early strength cement are
given in Table 2.6. (The ratio 1:3 represents the proportion of cement to aggregate by
mass.)

Table 2.6. Typical Values of Strength of a 1:3 Concrete made with Ultra High
Early Strength Portland Cement2.35

81
 More recent ultra high early strength cements have been reported2.12 to have a very
high C3S content, 60 per cent, and a very low C2S content, 10 per cent. The initial set
occurred at 70 minutes but the final set followed soon after, at 95 minutes.2.21 We
should note, however, that for the same mix proportions, the use of ultra high early
strength cement results in a lower workability.
Ultra high early strength cement has been used successfully in a number of
structures where early prestressing or putting into service is of importance. Shrinkage
and creep are not significantly different from those obtained with other Portland
cements when the mix proportions are the same;2.36 in the case of creep, the
comparison has to be made on the basis of the same stress/strength ratio (see p. 456).
The ultra high early strength cements discussed so far contain no integral
admixtures and are fundamentally of the Portland-cement-only variety. There exist
also cements with a proprietary composition. One of these is the so-called regulated-
set cement, or jet cement, developed in the United States. The cement consists
essentially of a mixture of Portland cement and calcium fluoroaluminate (C11A7.CaF2)
with an appropriate retarder (usually citric acid or lithium salts). The setting time of
the cement can vary between 1 and 30 minutes (the strength development being
slower the slower the setting) and is controlled in the manufacture of the cement as
the raw materials are interground and burnt together. Grinding is difficult because of
hardness differences.2.65
The early strength development is controlled by the content of calcium
fluoroaluminate: when this is 5 per cent, about 6 MPa (900 psi) can be achieved at 1
hour; a 50 per cent mixture will produce 20 MPa (3000 psi) at the same time or even
earlier. These values are based on a mix with a cement content of 330 kg/m3 (560
lb/yd3). The later strength development is similar to that of the parent Portland cement
but at room temperature there is virtually no gain in strength between 1 and 3 days.
A typical Japanese jet cement2.23 has a Blaine specific surface of 590 m2/kg and an
oxide composition (in per cent) as follows:

At a water/cement ratio of 0.30, compressive strengths of 8 MPa (1200 psi) at 2 hours,

and 15 MPa (2260 psi) at 6 hours, were reached.2.30 Drying shrinkage of concrete made
with jet cement was found2.23 to be lower than when Portland cement at the same
content per cubic metre of concrete was used. Also, permeability at ages up to 7 days
is very much lower.2.23 These features are important when regulated-set cement is used

82
for urgent repairs, for which this cement is particularly appropriate in view of its rapid
setting and very rapid early strength development. Clearly, the mixing procedure must
be appropriate. When required, a retarding admixture can be used.2.23 Regulated-set
cement is vulnerable to sulfate attack because of the high content of calcium
aluminate.2.37
There exist other special, very rapid-hardening cements. These are sold under
proprietary or trade names and have undisclosed composition. For these reasons, it
would not be appropriate or reliable to discuss them in this book. However, to give an
indication of what is available, at least in some countries, and to indicate the
performance of such cements, one of these will be discussed below. Let us call it
Cement X.
Cement X is a blended cement consisting of about 65 per cent of Portland cement
with a Blaine fineness of 500 m2/kg, about 25 per cent of Class C fly ash, and
undisclosed functional chemical additions. These are likely to include citric acid,
potassium carbonate and a superplasticizer, but no chlorides. The cement is used,
typically, at a content of 450 kg per cubic metre of concrete (or 750 lb/yd3) with a
water/cement ratio of approximately 0.25. The setting time is 30 minutes or more. It is
claimed that the concrete can be placed at temperatures slightly below freezing point,
but insulation of concrete to retain heat is necessary.
The strength development of concrete made with Cement X is very rapid: about 20
MPa (or 3000 psi) at 4 hours. The 28-day compressive strength is about 80 MPa (or
12 000 psi). The concrete is said to have a good resistance to sulfate attack and to
freezing and thawing, without air entrainment. The latter is due to the very low
water/cement ratio. Shrinkage is also said to be low.
These features make Cement X appropriate for rapid repair work, and possibly also
for precast concrete. It should be noted, however, that Cement X has an alkali content
of about 2.4 per cent (expressed as soda equivalent) and this should be borne in mind
when alkali-reactive aggregates may be used. Because of its high reactivity and
fineness, storage of this cement under very dry conditions is essential.
Low heat Portland cement
The rise in temperature in the interior of a large concrete mass due to the heat
development by the hydration of cement, coupled with a low thermal conductivity of
concrete, can lead to serious cracking (see p. 396). For this reason, it is necessary to
limit the rate of heat evolution of the cement used in this type of structure: a greater
proportion of the heat can then be dissipated and a lower rise in temperature results.
Cement having such a low rate of heat development was first produced for use in
large gravity dams in the United States, and is known as low heat Portland cement
(Type IV). However, for some time now, Type IV cement has not been produced in
the United States.
In the United Kingdom, low heat Portland cement is covered by BS 1370 : 1979,
which limits the heat of hydration of this cement to 250 J/g (60 cal/g) at the age of 7
days, and 290 J/g (70 cal/g) at 28 days.
The limits of lime content of low heat Portland cement, after correction for the
lime combined with SO3, are:

83
and

The rather lower content of the more rapidly hydrating compounds, C3S and C3A,
results in a slower development of strength of low heat cement as compared with
ordinary Portland cement, but the ultimate strength is unaffected. In any case, to
ensure a sufficient rate of gain of strength the specific surface of the cement must be
not less than 320 m2/kg. There is no separate recognition of low heat Portland cement
in the European Standard BS EN 197-1 : 2000.
In the United States, Portland–pozzolana cement Type P can be specified to be of
the low heat variety; the Type IP Portland–pozzolana cement can be required to have
moderate heat of hydration, which is denoted by the suffix MH. ASTM Standard C
595-10 deals with these cements.
In some applications, a very low early strength may be a disadvantage, and for this
reason a so-called modified (Type II) cement was developed in the United States. This
modified cement successfully combines a somewhat higher rate of heat development
than that of low heat cement with a rate of gain of strength similar to that of ordinary
Portland cement. Modified cement is recommended for use in structures where a
moderately low heat generation is desirable or where moderate sulfate attack may
occur. This cement is extensively used in the United States.
Modified cement, referred to as Type II cement, and low heat cement (Type IV)
are covered by ASTM C 150-09.
As mentioned earlier, Type IV cement has not been used in the United States for
some time, and the problem of avoiding excessive generation of heat due to the
hydration of cement is usually solved by other means. These include the use of fly ash
or pozzolana and a very low cement content. The cement used can be Type II cement
with a heat of hydration of 290 J/g (70 cal/g) at 7 days (offered as an option in ASTM
C 150-09), as compared with 250 J/g (60 cal/g) for Type IV cement.
Sulfate-resisting cement
In discussing the reactions of hydration of cement, and in particular the setting
process, mention was made of the reaction between C3A and gypsum (CaSO4.2H2O)
and of the consequent formation of calcium sulfoaluminate. In hardened cement,
calcium aluminate hydrate can react with a sulfate salt from outside the concrete in a
similar manner: the product of addition is calcium sulfoaluminate, forming within the
framework of the hydrated cement paste. Because the increase in the volume of the
solid phase is 227 per cent, gradual disintegration of concrete results. A second type
of reaction is that of base exchange between calcium hydroxide and the sulfates,
resulting in the formation of gypsum with an increase in the volume of the solid phase
of 124 per cent.
These reactions are known as sulfate attack. The salts particularly active are
magnesium sulfate and sodium sulfate. Sulfate attack is greatly accelerated if
accompanied by alternating wetting and drying.
The remedy lies in the use of cement with a low C3A content, and such cement is
known as sulfate-resisting Portland cement. The British Standard for this cement, BS
4027 : 1996, stipulates a maximum C3A content of 3.5 per cent. The SO3 content is
limited to 2.5 per cent. In other respects, sulfate-resisting cement is similar to ordinary

84
Portland cement but it is not separately recognized in BS EN 197-1 : 2000. In the
United States, sulfate-resisting cement is known as Type V cement and is covered by
ASTM C 150-09. This specification limits the C3A content to 5 per cent, and also
restricts the sum of the content of C4AF plus twice the C3A content to 25 per cent. The
magnesia content is limited to 6 per cent. There exists also a cement with moderate
sulfate resistance (ASTM C 595-10).
The role of C4AF is not quite clear. From the chemical standpoint, C4AF would be
expected to form calcium sulfoaluminate, as well as calcium sulfoferrite, and thus
cause expansion. It seems, however, that the action of calcium sulfate
on hydrated cement is smaller the lower the Al2O3:Fe2O3 ratio. Some solid solutions
are formed and they are liable to comparatively little attack. The tetracalcium ferrite is
even more resistant, and it may form a protective film over any free calcium
aluminate.2.6
As it is often not feasible to reduce the A12O3 content of the raw material,
Fe2O3 may be added to the mix so that the C4AF content increases at the expense of
C3A.2.7
An example of a cement with a very low Al2O3:Fe2O3 ratio is the Ferrari cement, in
whose manufacture iron oxide is substituted for some of the clay. A similar cement is
produced in Germany under the name of Erz cement. The name of iron-ore cement is
also used for this type of cement.
The low C3A content and comparatively low C4AF content of sulfate-resisting
cement mean that it has a high silicate content and this gives the cement a high
strength but, because C2S represents a high proportion of the silicates, the early
strength is low. The heat developed by sulfate-resisting cement is not much higher
than that of low heat cement. It could therefore be argued that sulfate-resisting cement
is theoretically an ideal cement but, because of the special requirements for the
composition of the raw materials used in its manufacture, sulfate-resisting cement
cannot be generally and cheaply made.
It should be noted that the use of sulfate-resisting cement may be disadvantageous
when there is a risk of the presence of chloride ions in the concrete containing steel
reinforcement or other embedded steel. The reason for this is that C3A binds chloride
ions, forming calcium chloroaluminate. In consequence, these ions are not available
for initiation of corrosion of the steel. This topic is discussed on p. 571.
Provision for a low-alkali sulfate-resisting cement is made in BS 4027 : 1996. In
this connection, it is worth noting that a low alkali content in cement is beneficial with
respect to sulfate attack, regardless of the C3A content in the cement. The reason for
this is that a low alkali content reduces the early availability of sulfate ions for
reaction with the C3A;2.12 it is not known whether this effect persists for a long time.
White cement and pigments
For architectural purposes, white or a pastel colour concrete is sometimes required. To
achieve best results it is advisable to use white cement with, of course, a suitable fine
aggregate and, if the surface is to be treated, also an appropriate coarse aggregate.
White cement has also the advantage that it is not liable to cause staining because it
has a low content of soluble alkalis.
White Portland cement is made from raw materials containing very little iron oxide
(less than 0.3 per cent by mass of clinker) and manganese oxide. China clay is
generally used, together with chalk or limestone, free from specified impurities. Oil or

85
gas is used as fuel for the kiln in order to avoid contamination by coal ash. Since iron
acts as a flux in clinkering, its absence necessitates higher kiln temperatures (up to
1650 °C) but sometimes cryolite (sodium aluminium fluoride) is added as a flux.
Contamination of the cement with iron during grinding of clinker has also to be
avoided. For this reason, instead of the usual ball mill, the rather inefficient flint
pebble grinding or expensive nickel and molybdenum alloy balls are used in a stone-
or ceramic-lined mill. The cost of grinding is thus higher, and this, coupled with the
more expensive raw materials, makes white cement rather expensive (about three
times the price of ordinary Portland cement).
Because of this, white cement concrete is often used in the form of a facing placed
against ordinary concrete backing, but great care is necessary to ensure full bond
between the two concretes. To obtain good colour, white concrete of rich-mix
proportions is generally used, the water/cement ratio being not higher than about 0.4.
A possible saving in some cases can be achieved by a partial replacement of white
cement by blastfurnace slag, which has a very light colour.
Strictly speaking, white cement has a faint green or yellow hue, depending on
impurities; traces of chromium, manganese, and iron are mainly responsible for the
slight coloration of green, bluish-green, and yellow, respectively.2.20
A typical compound composition of white Portland cement is given in Table
2.7 but the C3S and C2S contents may vary widely. White cement has a slightly lower
specific gravity than ordinary Portland cement, generally between 3.05 and 3.10.
Because the brightness of the white colour is increased by a higher fineness of cement,
it is usually ground to a fineness of 400 to 450 kg/m2. The strength of white Portland
cement is usually somewhat lower than that of ordinary Portland cement but white
cement nevertheless satisfies the requirements of BS 12 : 1996.

Table 2.7. Typical Compound Composition of White Portland Cement

White high-alumina cement is also made; this is considered on p. 103.

When a pastel colour is required, white concrete can be used as a base for painting.
Alternatively, pigments can be added to the mixer; those are powders of fineness
similar to, or higher than, that of cement. A wide range of colours is available; for
example, iron oxides can produce yellow, red, brown and black colours; chromic
oxide produces green colour, and titanium dioxide produces white colour.2.38 It is
essential that the pigments do not affect adversely the development of strength of the
cement or affect air entrainment. For instance, carbon black, which is extremely fine,
increases the water demand and reduces the air content of the mix. For this reason,
some pigments are marketed in the United States with an interground air-entraining
agent; it is, of course, essential to be aware of this at the mix proportioning stage.

86
 Mixing of concrete with pigments is not common because it is rather difficult to
maintain a uniform colour of the resulting concrete. An improvement in the dispersion
of the pigment can be obtained by the use of superplasticizers.2.42 However, it is
essential to verify the compatibility of any pigment with the admixtures which it is
proposed to use. When the mix contains silica fume, light-coloured pigment may not
perform well because of the extreme fineness of silica fume which exerts a masking
effect.
Requirements for pigments are given in BS EN 12878 : 2005; American
specification ASTM C 979-05 covers coloured and white pigments. It is desirable that
the 28-day compressive strength be not less than 90 per cent of the strength of a
pigment-free control mix, and the water demand is required to be not more than 110
per cent of the control mix. Setting time must not be unduly affected by the pigment.
It is essential that pigments are insoluble and not affected by light.
A better way to obtain a uniform and durable coloured concrete is to use coloured
cement. This consists of white cement interground with 2 to 10 per cent of pigment,
usually an inorganic oxide. Specifications for the use of this type of cement are given
by the individual manufacturers of this rather specialized product. Because the
pigment is not cementitious, slightly richer mixes than usual should be used. The use
of coloured concrete is reviewed by Lynsdale and Cabrera.2.38
For paving blocks, a ‘dry-shake’ of a mixture of pigment, cement, and hard fine
aggregate is sometimes applied prior to finishing.
Portland blastfurnace cement
Cements of this name consist of an intimate mixture of Portland cement and ground
granulated blastfurnace slag (in ASTM parlance, simply slag). This slag is a waste
product in the manufacture of pig iron, about 300 kg of slag being produced for each
tonne of pig iron. Chemically, slag is a mixture of lime, silica, and alumina, that is,
the same oxides that make up Portland cement but not in the same proportions. There
exist also non-ferrous slags; their use in concrete may become developed in the
future.2.39
Blastfurnace slag varies greatly in composition and physical structure depending
on the processes used and on the method of cooling of the slag. For use in the
manufacture of blastfurnace cement, the slag has to be quenched so that it solidifies as
glass, crystallization being largely prevented. This rapid cooling by water results also
in fragmentation of the material into a granulated form. Pelletizing, which requires
less water, can also be used.
Slag can make a cementitious material in different ways. Firstly, it can be used
together with limestone as a raw material for the conventional manufacture of
Portland cement in the dry process. Clinker made from these materials is often used
(together with slag) in the manufacture of Portland blastfurnace cement. This use of
slag, which need not be in glass form, is economically advantageous because lime is
present as CaO so that the energy to achieve decarbonation (see p. 3) is not required.
Secondly, granulated blastfurnace slag, ground to an appropriate fineness, can be
used on its own, but in the presence of an alkali activator or starter, as a cementitious
material; in other words, ground granulated blastfurnace slag, abbreviated as ggbs, is
a hydraulic material.2.41 It is used as such in masonry mortar and in other construction,
but the use of ggbs alone is outside the scope of this book.

87
 The third, and in most countries by far the major, use of ggbs is in Portland
blastfurnace cement, as defined in the opening paragraph of this section. This type of
cement can be produced either by intergrinding Portland cement clinker and dry
granulated blastfurnace slag (together with gypsum) or by dry blending of Portland
cement powder and ggbs. Both methods are used successfully, but it should be noted
that slag is harder than clinker, and this should be taken into account in the grinding
process. Separate grinding of granulated slag results in a smoother surface texture,
which is beneficial for workability.2.45
Another approach is to feed dry-ground granulated blastfurnace slag into the mixer
at the same time as Portland cement: Portland blastfurnace cement concrete is thus
manufactured in situ. This procedure is covered by BS 5328 : 1991 (withdrawn and
replaced by BS EN 206-1 : 2000).
A Belgian development is the Trief process in which wet-ground granulated slag is
fed in the form of a slurry direct into the concrete mixer, together with Portland
cement and aggregate. The cost of drying the slag is thus avoided, and grinding in the
wet state results in a greater fineness than would be obtained with dry grinding for the
same power input.
There are no detailed requirements for the content of the individual oxides in ggbs
to be used in concrete, but slags with the following percentages are known to be
satisfactory in cement:2.54

Lower amounts of lime and higher amounts of magnesia are also used.2.56 The
magnesia is not in crystalline form and does not therefore lead to harmful
expansion.2.58 Small amounts of iron oxide, manganese oxide, alkalis and sulfur can
also be present.
The specific gravity of ggbs is about 2.9, which is somewhat lower than the
specific gravity of Portland cement (that is, 3.15). The specific gravity of blended
cement is correspondingly affected.
When Portland blastfurnace cement is mixed with water, the Portland cement
component begins to hydrate first, although there is also a small amount of immediate
reaction of ggbs which releases calcium and aluminium ions into solution.2.56 The ggbs
then reacts with alkali hydroxide; this is followed by reaction with calcium hydroxide
released by Portland cement, C-S-H being formed.2.56
European Standard BS EN 197-1 : 2000 requires that, for use in the production of
any of the blended cements containing ggbs, the slag has to satisfy certain
requirements. According to BS 146 : 1996 and BS 4246 : 1996, at least two-thirds of
the slag must consist of glass. At least two-thirds of the total mass of slag must consist
of the sum of CaO, MgO, and SiO2. Also, the ratio of the mass of CaO plus MgO to
the mass of SiO2 must exceed 1.0. This ratio assures a high alkalinity, without which
the slag would be hydraulically inactive. The shape of ggbs is angular, in contrast to
fly ash. The replacement BS EN 197-4 : 2004 is not prescriptive in this respect.
The ASTM Specification C 989-09a prescribes a maximum proportion of 20 per
cent of ggbs coarser than a 45 μm sieve. The British Standards do not use such a

88
requirement. The specific surface of ggbs is not normally determined, but an increase
in fineness of Portland blastfurnace cement, accompanied by optimizing the
SO3 content, leads to an increased strength; when the specific surface is increased
from 250 to 500 m2/kg (by the Blaine method) the strength is more than doubled.2.59
The American approach, given in ASTM C 989-09a, is to grade blastfurnace slag
according to its hydraulic activity. This is determined by the strength of mortars of
standard mass proportions containing slag as compared with mortars containing
Portland cement only. Three grades are recognized.
European Standard BS EN 197-1 : 2000 recognizes three classes of Portland
blastfurnace cement, called Blastfurnace cement III/A, III/B, and III/C. All of them
are allowed to contain up to 5 per cent of filler, but they differ in the mass of ggbs as a
percentage of the mass of the total cementitious material, that is Portland cement plus
ggbs exclusive of calcium sulfate and the manufacturing additive. The percentages of
slag are as follows:

Class III/C Blastfurnace cement, at its upper limit of ggbs, is virtually a pure slag
cement, which, as already stated, will not be further considered in this book.
Cements with a high content of ggbs can be used as low heat cements in structures
in which a large mass of concrete is to be placed so that the temperature increase
arising from the early development of the heat of hydration of cement needs to be
controlled; this topic is considered on p. 395. British Standard BS 4246 : 2002
(replaced by BS EN 197-4 : 2004) provides an option for a purchaser’s specification
of the heat of hydration. It must not be forgotten that a concomitant of a low rate of
heat development is a low rate gain of strength. Therefore, in cold weather the low
heat of hydration of Portland blastfurnace cement, coupled with a moderately low rate
of strength development, can lead to frost damage.
Cements containing ggbs are often also beneficial from the standpoint of resistance
to chemical attack. This is discussed on p. 667.
Hydraulic activity of ggbs is conditional on its high fineness but, as in the case of
other cements, the fineness of Portland blastfurnace cements is not specified in the
British Standards. The only exception is when ggbs and Portland cement are dry
blended: in that case, ggbs has to conform to BS 6699 : 1992 (1998). In practice, the
fineness of ggbs tends to be higher than that of Portland cement.
In addition to the Portland blastfurnace cements discussed above, BS EN 197-1 :
2000 recognizes two cements containing lesser amounts of slag. These are cements
Class II A-S with 6 to 20 per cent of ggbs and Class II B-S with 21 to 35 per cent of
ggbs, by mass. These are called Portland slag cements; they form part of the large
variety of Class II cements, all of which consist predominantly of Portland cement,
but are blended with another cementitious material (see Table 2.4).
British Standards BS 146 : 1996 and BS 4246 : 1996 contain some additional
requirements and also classify the cements on the basis of compressive strength. The
classification is the same as for other cements but it is important to note that two of
the classes of Portland blastfurnace cement are subdivided into categories: a low early
strength; an ordinary early strength; and a high early strength. These are a reflection
of the progress of hydration of blastfurnace cements: at very early ages, the rate of

89
hydration is lower than in the case of Portland cement alone. British Standard BS
4246 : 1996 allows cements with a slag content of 50 to 85 per cent by mass to have a
7-day compressive strength as low as 12 MPa.
Supersulfated cement
Supersulfated cement is made by intergrinding a mixture of 80 to 85 per cent of
granulated blastfurnace slag with 10 to 15 per cent of calcium sulfate (in the form of
dead-burnt gypsum or anhydrite) and up to 5 per cent of Portland cement clinker. A
fineness of 400 to 500 m2/kg is usual. Supersulfated cement is thus fundamentally
different from Portland cement, in which calcium silicate is the main component. The
cement has to be stored under very dry conditions as otherwise it deteriorates rapidly.
Supersulfated cement is used extensively in Belgium, where it is known as ciment
métallurgique sursulfaté, also in France, and was previously manufactured in
Germany (under the name of Sulfathüttenzement). In the United Kingdom, the cement
was covered by BS 4248- : 2004 (withdrawn) but, because of production difficulties,
the manufacture of the cement has been discontinued. The European standard for
supersulfated cement is BS EN 15743 : 2010, which gives physical and chemical
requirements.
Supersulfated cement is highly resistant to sea water and can withstand the highest
concentrations of sulfates normally found in soil or ground water, and is also resistant
to peaty acids and to oils. Concrete with a water/cement ratio not greater than 0.45 has
been found not to deteriorate in contact with weak solutions of mineral acids of pH
down to 3.5. For these reasons, supersulfated cement is used in the construction of
sewers and in contaminated ground, although it has been suggested that this cement is
less resistant than sulfate-resisting Portland cement when the sulfate concentration
exceeds 1 per cent.2.31
The heat of hydration of supersulfated cement is very low: about 170 to 190 J/g
(40 to 45 cal/g) at 7 days, and 190 to 210 J/g (45 to 50 cal/g) at 28 days.2.6 The cement
is, therefore, suitable for mass concrete construction but care must be taken if used in
cold weather because the rate of strength development is considerably reduced at low
temperatures. The rate of hardening of supersulfated cement increases with
temperature up to about 50 °C (122 °F), but at higher temperatures anomalous
behaviour has been encountered. For this reason, steam curing above 50 °C (122 °F)
should not be used without prior tests. It may also be noted that supersulfated cement
should not be mixed with Portland cements because the lime released by the hydration
of an excessive amount of the latter interferes with the reaction between the slag and
the calcium sulfate.
Wet curing for not less than four days after casting is essential as premature drying
out results in a friable or powdery surface layer, especially in hot weather, but the
depth of this layer does not increase with time.
Supersulfated cement combines chemically with more water than is required for
the hydration of Portland cement, so that concrete with a water/cement ratio of less
than 0.4 should not be made. Mixes leaner than about 1:6 are not recommended. The
decrease in strength with an increase in the water/cement ratio has been reported to be
smaller than in other cements but, because the early strength development depends on
the type of slag used in the manufacture of the cement, it is advisable to determine the
actual strength characteristics prior to use. Typical strengths attainable are given

90
in Table 2.8. BS EN 15743 : 2010 specifies three classes of strength: 32.5, 42.5, and
52.5.

Table 2.8. Typical Values of Strength of Supersulfated Cement2.6

Pozzolanas
One of the common materials classified as cementitious in this book (although in
reality only in latent form) is pozzolana, which is a natural or artificial material
containing silica in a reactive form. A more formal definition of ASTM 618-08a
describes pozzolana as a siliceous or siliceous and aluminous material which in itself
possesses little or no cementitious value but will, in finely divided form and in the
presence of moisture, chemically react with calcium hydroxide at ordinary
temperatures to form compounds possessing cementitious properties. It is essential
that pozzolana be in a finely divided state as it is only then that silica can combine
with calcium hydroxide (produced by the hydrating Portland cement) in the presence
of water to form stable calcium silicates which have cementitious properties. We
should note that the silica has to be amorphous, that is, glassy, because crystalline
silica has very low reactivity. The glass content can be determined by X-ray
diffraction spectroscopy or by solution in hydrochloric acid and potassium
hydroxide.2.24
Broadly speaking, pozzolanic materials can be natural in origin or artificial. The
main artificial pozzolanic material, fly ash, will be considered in the next section.
The natural pozzolanic materials most commonly met with are: volcanic ash – the
original pozzolana – pumicite, opaline shales and cherts, calcined diatomaceous earth,
and burnt clay. ASTM C 618-08a describes these materials as Class N.
Some natural pozzolanas may create problems because of their physical properties;
e.g. diatomaceous earth, because of its angular and porous form, requires a high water
content. Certain natural pozzolanas improve their activity by calcination in the range
of 550 to 1100 °C, depending on the material.2.63
Rice husks are a natural waste product and there is interest in using this material in
concrete. Rice husks have a very high silica content, and slow firing at a temperature
of 500 to 700 °C results in an amorphous material with a porous structure. Thus the
specific surface (measured by nitrogen adsorption) can be as high as 50 000 m2/kg,
even though the particle size is large: 10 to 75 μm.2.26 The rice husk ash particles have

91
complex shapes, reflecting their plant origins2.27 and they therefore have a high water
demand unless interground with clinker so as to break down the porous structure.
Rice husk ash is reported to contribute to the strength of concrete already at 1 to 3
days.2.26
However, to achieve adequate workability, as well as high strength, the use of
superplasticizers may be necessary;2.28 this negates the economic benefits of the use of
rice husk ash in less affluent areas of the world where collection of the husks for
processing may also present problems. The use of rice husks can lead to increased
shrinkage2.80 but this has not been confirmed.
There exist also other processed amorphous silica materials. One of these
is metakaolin, obtained by calcination of pure or refined kaolinitic clay at a
temperature of between 650 and 850 °C, followed by grinding to achieve a fineness of
700 to 900 m2/kg. The resulting material exhibits high pozzolanicity.2.53,2.60
The use of siliceous clay, ground to a very high fineness (specific surface of 4000
to 12 000 m2/kg, determined by nitrogen adsorption), as a highly reactive pozzolana
has been suggested by Kohno et al.2.61
For an assessement of pozzolanic activity with cement, ASTM C 311-07
prescribes the measurement of a strength activity index. This is established by the
determination of strength of mortar with a specified replacement of cement by
pozzolana. The outcome of the test is influenced by the cement used, especially its
fineness and alkali content.2.25 There is also a pozzolanic activity index with lime,
which determines the total activity of pozzolana.
The pozzolanicity of pozzolanic cements, that is, cements containing between 11
and 55 per cent of pozzolana and silica fume according to BS EN 197-1 : 2000, is
tested according to BS EN 196-5 : 2005. The test compares the quantity of calcium
hydroxide in an aqueous solution in contact with the hydrated pozzolanic cement,
with the quantity of calcium hydroxide which saturates a solution of the same
alkalinity. If the former concentration is lower than the latter, then the pozzolanicity
of the cement is considered to be satisfactory. The underlying principle is that the
pozzolanic activity consists of fixing of calcium hydroxide by the pozzolana so that
the lower the resulting quantity of calcium hydroxide the higher the pozzolanicity.
Pozzolanicity is still imperfectly understood; specific surface and chemical
composition are known to play an important role but, because they are inter-related,
the problem is complex. It has been suggested that, in addition to reacting with
Ca(OH)2, pozzolanas react also with C3A or its products of hydration.2.76 A good
review of the subject of pozzolanicity has been written by Massazza and Costa.2.77
There exists one other material, silica fume, which is formally an artificial
pozzolana but whose properties put it into a class of its own. For this reason, silica
fume will be considered in a separate section (see p. 86).
Fly ash
Fly ash, known also as pulverized-fuel ash, is the ash precipitated electrostatically or
mechnically from the exhaust gases of coal-fired power stations; it is the most
common artificial pozzolana. The fly ash particles are spherical (which is
advantageous from the water requirement point of view) and have a very high
fineness: the vast majority of particles have a diameter between less than 1 μm and
100 μm, and the specific surface of fly ash is usually between 250 and 600 m2/kg,

92
using the Blaine method. The high specific surface of the fly ash means that the
material is readily available for reaction with calcium hydroxide.
The specific surface of fly ash is not easy to determine because, in the air
permeability test, the spherical particles pack more closely than the irregularly shaped
particles of cement so that the resistance of fly ash to air flow is greater. On the other
hand, the porous carbon particles in the ash allow air to flow through them, leading to
a misleadingly high air flow.2.62 Moreover, the determination of the specific gravity of
fly ash (which enters the calculation of the specific surface, see p. 22) is affected by
the presence of hollow spheres (whose specific gravity can be less than 1).2.62 At the
other extreme, some small particles which contain magnetite or haematite have a high
specific gravity. The typical overall value of specific gravity is 2.35. An important use
of the determination of the specific surface of fly ash is in detecting its variability.2.64
The American classification of fly ash, given in ASTM C 618-08a, is based on the
type of coal from which the ash originates. The most common fly ash derives from
bituminous coal, is mainly siliceous, and is known as Class F fly ash.
Sub-bituminous coal and lignite result in high-lime ash, known as Class C fly ash.
This will be considered later in the present section.
The pozzolanic activity of Class F fly ash is in no doubt, but it is essential that it
has a constant fineness and a constant carbon content. The two are often
interdependent because the carbon particles tend to be coarser. Modern boiler plants
produce fly ash with a carbon content of about 3 per cent, but much higher values are
encountered in fly ash from older plants. The carbon content is assumed to be equal to
the loss on ignition, although the latter includes also any combined water or fixed
CO2 present.2.64 British Standard BS EN 450-1 : 2005 1 : 1997 specifies a maximum 12
per cent residue on the 45 μm sieve, which is a convenient basis of classification of
size.
The main requirements of ASTM C 618-08a are: a minimum content of 70 per
cent of silica, alumina, and ferric oxide taken all together, a maximum SO3 content of
5 per cent, a maximum loss on ignition of 12 per cent. Also, to control any alkali-
aggregate reaction, the expansion of a mix with fly ash should not exceed that of a
low alkali cement control mix at 14 days. British Standard BS 3892-1 : 1997 specifies
a maximum content of SO3 of 2.5 per cent and some other requirements. A limitation
on the MgO content is no longer specified because it exists in a non-reactive form.
It should be noted that fly ash may affect the colour of the resulting concrete, the
carbon in the ash making it darker. This may be of importance from the standpoint of
appearance, especially when concretes with and without fly ash are placed side by
side.
Let us now turn to Class C fly ash, that is, high-lime ash originating from lignite
coal. Such ash may occasionally have a lime content as high as 24 per cent.2.63 High-
lime ash has some cementitious (hydraulic) properties of its own but, because its lime
will combine with the silica and alumina portions of the ash, there will be less of these
compounds to react with the lime liberated by the hydration of cement. The carbon
content is low, the fineness is high, and the colour is light. However, the MgO content
can be high, and some of the MgO as well as some of the lime can lead to deleterious
expansion.2.63
The behaviour of high-lime ash is sensitive to temperature: specifically, in mass
concrete when a rise in temperature occurs, the products of reaction may not be of
high strength. However, the development of strength is not simply related to

93
temperature, being satisfactory in the region of 120 to 150 °C (250 to 300 °F) but not
at about 200 °C (about 400 °F) when the products of reaction are substantially
different.2.55
Pozzolanic cements
Pozzolanas, being a latent hydraulic material, are always used in conjunction with
Portland cement. The two materials may be interground or blended. Sometimes, they
can be combined in the concrete mixer. The possibilities are thus similar to those of
granulated blastfurnace slag (see p. 79). By far the largest proportion of pozzolanas
used consists of siliceous fly ash (Class F), and we shall concentrate on that material.
European Standard BS EN 197-1 : 2000 recognizes two subclasses of Portland fly
ash cement: Class II/A-V with a fly ash content of 6 to 20 per cent, and Class II/B-V
with a fly ash content of 21 to 35 per cent. The British Standard for Portland
pulverized-fuel ash cements, BS 6588 : 1996, has somewhat different limits for the fly
ash content, the maximum value being 40 per cent. There is no great significance in
the precise upper limit on the fly ash content. However, BS 6610 : 1991 allows an
even higher content of fly ash, namely 53 per cent, in so-called pozzolanic cement.
Like the high slag blastfurnace cement (see p. 81), pozzolanic cement has a low 7-day
strength (minimum of 12 MPa) but also a low 28-day strength: minimum of 22.5 MPa.
The concomitant advantage is a low rate of heat development so that pozzolanic
cement is a low heat cement. Additionally, pozzolanic cement has some resistance to
sulfate attack and to attack by weak acids.
Silica fume
Silica fume is a relatively recent arrival among cementitious materials. It was
originally introduced as a pozzolana. However, its action in concrete is not only that
of a very reactive pozzolana but is also beneficial in other respects (see p. 669). It can
be added that silica fume is expensive.
Silica fume is also referred to as microsilica or condensed silica fume, but the term
‘silica fume’ has become generally accepted. It is a by-product of the manufacture of
silicon and ferrosilicon alloys from high-purity quartz and coal in a submerged-arc
electric furnace. The escaping gaseous SiO oxidizes and condenses in the form of
extremely fine spherical particles of amorphous silica (SiO2); hence, the name silica
fume. Silica in the form of glass (amorphous) is highly reactive, and the smallness of
the particles speeds up the reaction with calcium hydroxide produced by the hydration
of Portland cement. The very small particles of silica fume can enter the space
between the particles of cement, and thus improve packing. When the furnace has an
efficient heat recovery system, most of the carbon is burnt so that silica fume is
virtually free from carbon and is light in colour. Furnaces without a full heat recovery
system leave some carbon in the fume, which is therefore dark in colour.
The production of silicon alloys, which include non-ferrous metals, such as
ferrochromium, ferromanganese, and ferromagnesium, also results in the formation of
silica fume but its suitability for use in concrete has not yet been established.2.67
The usual ferrosilicon alloys have nominal silicon contents of 50, 75, and 90 per
cent; at 48 per cent, the product is called silicon metal. The higher the silicon content
in the alloy the higher the silica content in the resulting silica fume. Because the same
furnace can produce different alloys, it is important to know the provenance of any
silica fume to be used in concrete. In particular, ferrosilicon with a 50-per cent content
of silicon results in a silica fume with a content of silica of only about 80 per cent.

94
However, steady production of a given alloy results in a silica fume with constant
properties.2.66 Typical silica contents are as follows (per cent): silicon metal, 94 to 98;
90 per cent ferrosilicon, 90 to 96; and 75 per cent ferrosilicon, 86 to 90.2.66
The specific gravity of silica fume is generally 2.20, but it is very slightly higher
when the silica content is lower.2.66 This value can be compared with the specific
gravity of Portland cement, which is 3.15. The particles of silica fume are extremely
fine, most of them having a diameter ranging between 0.03 and 0.3 μm; the median
diameter is typically below 0.1 μm. The specific surface of such fine particles cannot
be determined using the Blaine method; nitrogen adsorption indicates a specific
surface of about 20 000 m2/kg, which is 13 to 20 times higher than the specific surface
of other pozzolanic materials, determined by the same method.
Such a fine material as silica fume has a very low bulk density: 200 to 300
kg/m3 (12 to 19 lb/ft3). Handling this light powder is difficult and expensive. For this
reason, silica fume is available in the densified form of micropellets, that is,
agglomerates of the individual particles (produced by aeration), with a bulk density of
500 to 700 kg/m3. Another form of silica fume is a slurry of equal parts by mass of
water and silica fume. The density of the slurry is about 1300 to 1400 kg/m3. The
slurry is stabilized and has been reported to have a pH of about 5.5, but this is of no
consequence with respect to the use in concrete.2.68 Periodic agitation is necessary to
maintain a uniform distribution of the silica fume in the slurry. Admixtures, such as
water reducers, superplasticizers or retarders, can be included in the slurry.2.69
Each of the different forms in which silica fume is available has operational
advantages, but all forms can be successfully used; claims of significant beneficial
effects of one or other of these forms upon the resulting concrete have not been
substantiated.2.70
Although silica fume is usually incorporated in the mix at the batcher, in some
countries, blended cement containing silica fume, usually 6.5 to 8 per cent by mass, is
produced.2.71 Such a blended cement simplifies the batching operations but, obviously,
the content of silica fume in the total cementitious material cannot be varied to suit
specific needs.
Few standards for silica fume or its use in concrete are in existence. ASTM C
1240-05 specifies the requirements for silica fume, but ASTM C 618-08a, by its title,
excludes it. Indeed, the clause about the water requirement in that standard may well
not be satisfied by silica fume.
Fillers
In the classification of blended Portland cements (see p. 65) it was mentioned that
fillers may be included up to a certain maximum content. Indeed, fillers have been
used in many countries for some time but it is only recently that their use became
permitted in the United Kingdom.
A filler is a very finely-ground material, of about the same fineness as Portland
cement, which, owing to its physical properties, has a beneficial effect on some
properties of concrete, such as workability, density, permeability, capillarity, bleeding,
or cracking tendency. Fillers are usually chemically inert but there is no disadvantage
if they have some hydraulic properties or if they enter into harmless reactions with the
products of reaction in the hydrated cement paste. Indeed, it has been found by
Zielinska2.44 that CaCO3, which is a common filler, reacts with C3A and C4AF to
produce 3CaO.Al2O3.CaCO3.11H2O.

95
 Fillers can enhance the hydration of Portland cement by acting as nucleation sites.
This effect was observed in concrete containing fly ash and titanium dioxide in the
form of particles smaller than 1 μm.2.72 Ramachandran2.74 found that, in addition to its
nucleation role in the hydration of cement, CaCO3 becomes partly incorporated into
the C-S-H phase. This effect on the structure of the hydrated cement paste is
beneficial.
Fillers can be naturally occurring materials or processed inorganic mineral
materials. What is essential is that they have uniform properties, and especially
fineness. They must not increase the water demand when used in concrete, unless
used with a water-reducing admixture, or adversely affect the resistance of concrete to
weathering or the protection against corrosion which concrete provides to the
reinforcement. Clearly, they must not lead to a long-term retrogression of strength of
concrete, but such a problem has not been encountered.
Because the action of fillers is predominantly physical, they have to be physically
compatible with the cement in which they are included. Because the filler is softer
than clinker, it is necessary to grind the composite material longer so as to ensure the
presence of some very fine cement particles, which are necessary for early strength.
Although BS EN 197-1 : 2000 limits the filler content to 5 per cent, it allows the
use of limestone up to 35 per cent, provided the remaining cementitious material is
Portland cement only. This cement is known as Portland limestone cement (Class
II/B-L). As limestone is in effect a type of filler, the limestone cement can be said to
have a filler content of up to 35 per cent. It can be expected that, for some purposes,
blended cements with a filler content of 15, or even 20, per cent are likely to be
popular in the future. A higher filler content may cause a reduction in the strength of
concrete: 10 per cent for a filler content of 10 per cent, and 12 per cent for 20 per
cent.1.97 These losses can be compensated for by a reduced ration of water to total
cementitious material (w/cm).
Other cements
Among the numerous cements developed for special uses, anti-bacterial cement is of
interest. It is a Portland cement interground with an anti-bacterial agent which
prevents microbiological fermentation. This bacterial action is encountered in
concrete floors of food processing plants where the leaching out of cement by acids is
followed by fermentation caused by bacteria in the presence of moisture. Anti-
bacterial cement can also be successfully used in swimming pools and similar places
where bacteria or fungi are present.
Another special cement is the so-called hydrophobic cement, which deteriorates
very little during prolonged storage under unfavourable conditions. This cement is
obtained by intergrinding Portland cement with 0.1 to 0.4 per cent of oleic acid.
Stearic acid or pentachlorophenol can also be used.2.10 These additions increase the
grindability of clinker, probably due to electrostatic forces resulting from a polar
orientation of the acid molecules on the surface of the cement particles. Oleic acid
reacts with alkalis in cement to form calcium and sodium oleates which foam, so that
air-entraining results. When this is not desired a detraining agent, such as tri-n-butyl
phosphate, has to be added during grinding.2.11
The hydrophobic properties are due to the formation of a water-repellent film
around each particle of cement. This film is broken during the mixing of the concrete,
and normal hydration takes place but early strength is rather low.

96
 Hydrophobic cement is similar in appearance to ordinary Portland cement but has
a characteristic musty smell. In handling, the cement seems more flowing than other
Portland cements.
Masonry cement, used in mortar in brickwork, is made by integrinding Portland
cement, limestone and an air-entraining agent, or alternatively Portland cement and
hydrated lime, granulated slag or an inert filler, and an air-entraining agent; other
ingredients are usually also present. Masonry cements make a more plastic mortar
than ordinary Portland cement; they also have a greater water-retaining property and
lead to lower shrinkage. The strength of masonry cements is lower than that of
ordinary Portland cement, particularly because a high air content is introduced, but
this low strength is generally an advantage in brick construction. Masonry cement
must not be used in structural concrete. The specification for masonry cement is given
in ASTM C 91-05.
Three further cements should be mentioned. One is expansive cement, which has
the property of expanding in its early life so as to counteract contraction induced by
drying shrinkage. For this reason, expanding cement will be considered in Chapter 9.
The second cement is oil-well cement. This is a highly specialized product, based
on Portland cement, used for grout or slurry to be pumped to depths of up to
thousands of metres in the earth’s crust where temperature can exceed 150 °C (or
300 °F) and pressure can be 100 MPa (or 15 000 psi). These values would apply
typically to depths of about 5000 m (or 16 000 ft) but exploration holes to a depth of
10 000 m (or 33 000 ft) have been drilled and grouted.
The cements to be used in grout under these conditions must not set before
reaching distant locations but subsequently they have to gain strength rapidly so as to
allow resumption of the drilling operations. Sulfate resistance is often also required.
Several classes of oil-well cement are recognized by the American Petroleum Institute,
which prepares specifications for oil-well cements.2.21
Essentially, oil-well cements have to have certain special features: (a) to have a
particular fineness (to ‘hold’ a large amount of water); (b) to contain retarders or
accelerators (see Chapter 5); (c) to contain friction reducers (to improve flow); (d) to
contain lightweight additives (such as bentonite) to lower the density of the grout or
densifying additives (such as barytes or haematite) to increase the density of the grout;
and (e) to contain pozzolana or silica fume (to improve the strength at high
temperatures).
Finally, we should mention natural cement. This is the name given to a cement
obtained by calcining and grinding a so-called cement rock, which is a clayey
limestone containing up to 25 per cent of argillaceous material. The resulting cement
is similar to Portland cement, and is really intermediate between Portland cement and
hydraulic lime. Because natural cement is calcined at temperatures too low for
sintering, it contains practically no C3S and is therefore slow hardening. Natural
cements are rather variable in quality as adjustment of composition by blending is not
possible. Because of this, as well as for economic reasons, natural cements are
nowadays rarely used.
Which cement to use
The wide variety of cement Types (in American nomenclature) and cement Classes
(in European classification) and, above all, of cementitious and other materials used in

97
blended cements, may result in a bewildering impression. Which cement is best?
Which cement should be used for a given purpose?
There is no simple answer to these questions but a rational approach will lead to
satisfactory solutions.
First of all, no single cement is the best one under all circumstances. Even if cost is
ignored, pure Portland cement is not the all-round winner, although in the past
commercial interests extolled it as the true unadulterated product, second to none. As
far back as 1985, about one-half of all cement produced in Western Europe and in
China was blended, about two-thirds in India and in what was the Soviet Union, but
only a minimal proportion in North America and in the United Kingdom,2.29 possibly
because of the influence of the Portland cement lobby there.
The use of blended cements has been steadily increasing in the 1980s and the
1990s, and it can be confidently expected that blended cements will eventually form
the bulk of cements used world-wide. In the words of Dutron,2.29 “pure Portland
cements will be regarded as special cements reserved for applications where
exceptional performance is required, particularly as far as mechanical strength is
concerned”. Even this last caveat is no longer valid as high performance concrete is
best made with blended cements. Moreover, the durability of blended cements is
equal to, and often better than, that of pure Portland cement.
So, if no single cement is the best all round, we should look at the question: which
cement should be used for a given purpose?
The chapters which follow discuss the properties of concrete both in the fresh state
and when hardened. Many of these properties depend, to a greater or lesser extent, on
the properties of the cement used: it is on this basis that the choice of cement can be
made. However, in many cases, no one cement is the best one: more than one Type or
Class can be used. The choice depends on availability, on cost – that important
element in engineering decision-making – and on the particular circumstances of
equipment, skilled labour force, speed of construction and, of course, on the
exigencies of the structure and its environment.
It is intended to refer to the relevant properties of the different cements in the
chapters dealing with fresh concrete, strength and, especially, durability, and also
in Chapter 13, dealing with concretes with particular properties. Thus, it is there that
views on the choice or appropriateness of various cements can be found.
High-alumina cement
The search for a solution to the problem of attack by gypsum-bearing waters on
Portland cement concrete structures in France led Jules Bied to the development of a
high-alumina cement, at the beginning of the twentieth century. This cement is very
different in its composition, and also in some properties, from Portland cement and
Portland blended cements so that its structural use is severely limited, but the
concreting techniques are similar. For full treatment of the topic, the reader may
consult a specialized book.*
*
A. M. Neville, in collaboration with P. J. Wainwright, High-alumina Cement
Concrete (Construction Press, Longman Group, 1975).
Manufacture
From the name of the cement – high-alumina – it can be inferred that it contains a
large proportion of alumina: typically, about 40 per cent each of alumina and lime,

98
with about 15 per cent of ferrous and ferric oxides, and about 5 per cent of silica.
Small amounts of TiO2, magnesia, and the alkalis can also be present.
The raw materials are usually limestone and bauxite. Bauxite is a residual deposit
formed by the weathering, under tropical conditions, of rocks containing aluminium,
and consists of hydrated alumina, oxides of iron and titanium, and small amounts of
silica.
There are several processes of manufacture of high-alumina cement. In one
process, bauxite is crushed into lumps not larger than 100 mm (or 4 in.). Dust and
small particles formed during this fragmentation are cemented into briquettes of
similar size because dust would tend to damp the furnace. The second main raw
material is usually limestone, also crushed to lumps of about 100 mm (or 4 in.).
Limestone and bauxite, in the required proportions, are fed into the top of a
furnace which is a combination of the cupola (vertical stack) and reverberatory
(horizontal) types. Pulverized coal is used for firing, its quantity being about 22 per
cent of the mass of the cement produced. In the furnace, the moisture and carbon
dioxide are driven off and the materials are heated by the furnace gases to the point of
fusion at about 1600 °C. The fusion takes place at the lower end of the stack so that
the molten material falls into the reverberatory furnace and thence through a spout
into steel pans. The melt is now solidified into pigs, fragmented in a rotary cooler, and
then ground in a tube mill. A very dark grey powder with a fineness of 290 to 350
m2/kg is produced.
Because of the high hardness of high-alumina cement clinker, the power
consumption and the wear of tube mills are considerable. This, coupled with the high
prime cost of bauxite and the high temperature of firing, leads to a high price of high-
alumina cement, compared with Portland cement. The price is, however, compensated
for by some valuable properties for specific purposes.
It may be noted that, unlike the case of Portland cement, the materials used in the
manufacture of high-alumina cement are completely fused in the kiln. This fact gave
rise to the French name ciment fondu, and ‘fondu cement’ is sometimes used as a
colloquial name in English.
Because of adverse publicity, associated with high-alumina cement in the United
Kingdom in the 1970s (see p. 99), there have been attempts to use an alternative name
of aluminous cement. However, this name is not correct because other cements, such
as supersulfated cement and slag cements, also contain alumina in significant
proportions. Yet a third name, calcium aluminate cement, is more appropriate but then,
by contrast, we should refer to Portland cement as calcium silicate cement; this
appellation is never used. In this book, we shall therefore use the traditional name of
high-alumina cement.
High-alumina cement is no longer manufactured in the United Kingdom. However,
there exists a British Standard for high alumina cement, BS 915 : 1972 (1983), which
refers to BS 4550-3.1 : 1978 for fineness, strength, setting time and soundness. There
is a European standard BS EN 14647 : 2005.
Resistance to chemical attack
As mentioned earlier, high-alumina cement was first developed to resist sulfate attack,
and it is indeed highly satisfactory in this respect. This resistance to sulfates is due to
the absence of Ca(OH)2 in hydrated high-alumina cement and also to the protective
influence of the relatively inert alumina gel formed during hydration.2.16 However, lean

99
mixes are very much less resistant to sulfates.2.6 Also, the chemical resistance
decreases drastically after conversion (see p. 95).
High-alumina cement is not attacked by CO2 dissolved in pure water. The cement
is not acid-resisting but it can withstand tolerably well very dilute solutions of acids
(pH greater than about 4) found in industrial effluents, but not of hydrochloric,
hydrofluoric or nitric acids. On the other hand, caustic alkalis, even in dilute solutions,
attack high-alumina cement with great vigour by dissolving the alumina gel. The
alkalis may have their origin outside (e.g. by percolation through Portland cement
concrete) or in the aggregate. The behaviour of this cement in the presence of many
agents has been studied by Hussey and Robson.2.16
It may be noted that, although high-alumina cement stands up extremely well to
sea water, this water should not be used as mixing water; the setting and hardening of
the cement are adversely affected, possibly because of the formation of
chloroaluminates. Likewise, calcium chloride must never be added to high-alumina
cement.
Physical properties of high-alumina cement
A feature of high-alumina cement is its very high rate of strength development. About
80 per cent of its ultimate strength is achieved at the age of 24 hours, and even at 6 to
8 hours the concrete is strong enough for the side formwork to be struck and for the
preparation for further concreting to take place. Concrete made with high-alumina
cement, at a content of 400 kg/m3 (or 680 lb/yd3) and a water/cement ratio of 0.40, at
25 °C (77 °F), can reach a compressive strength (measured on cubes) of about 30
MPa (or 4500 psi) at 6 hours, and more than 40 MPa (or 6000 psi) at 24 hours. The
high rate of gain of strength is due to the rapid hydration, which in turn means a high
rate of heat development. This can be as high as 38 J/g per hour (9 cal/g per hour)
whereas for rapid-hardening Portland cement the rate is never higher than 15 J/g per
hour (3.5 cal/g per hour). However, the total heat of hydration is about the same for
both types of cement.
It should be stressed that the rapidity of hardening is not accompanied by rapid
setting. In fact, high-alumina cement is slow setting but the final set follows the initial
set more rapidly than is the case in Portland cement. Typical values for high-alumina
cement are: initial set at hours, and final set, 30 minutes later. Of the compounds
present in the high-alumina cement, C12A7 sets in a few minutes, whereas CA is
considerably more slow-setting, so that the higher the C: A ratio in the cement the
more rapid the set. On the other hand, the higher the glass content of the cement the
slower the set. It is likely that, because of its rapid setting properties, C12A7 is
responsible for the loss of workability of many high-alumina cement concretes, which
takes place within 15 or 20 minutes of mixing. Temperatures between 18 and 30 °C
(64 to 86 °F) slow down the setting but, above about 30 °C (86 °F), the setting is
rapidly accelerated; the reasons for this anomalous behaviour are not clear.2.40
The setting time of high-alumina cement is greatly affected by the addition of
plaster, lime, Portland cement and organic matter and for this reason no additives
should be used.
In the case of Portland cement–high-alumina cement mixtures, when either cement
constitutes between 20 and 80 per cent of the mixture, flash set may occur. Typical
data2.81 are shown in Fig. 2.4 but actual values vary for different cements, and trial tests
should be made with any given cements. When the Portland cement content is low,

100
the accelerated setting is due to the formation of a hydrate of C4A by the addition of
lime from the Portland cement to calcium aluminate from the high-alumina cement.
When the high-alumina cement content is low, gypsum contained in the Portland
cement reacts with hydrated calcium aluminates, and as a consequence the now non-
retarded Portland cement may exhibit a flash set.

Fig. 2.4. Setting time of Portland-high-alumina cement mixtures2.81

Mixtures of the two cements in suitable proportions are used when rapid setting is
of vital importance, e.g. for stopping the ingress of water, or for temporary
construction between the tides, but the ultimate strength of such pastes is quite low
except when the high-alumina cement content is very high. However, the use of high-
alumina cement for the purpose of shortening the setting time of concrete made with
Portland cement is discouraged by ACI 517.2R-87 (Revised 1991).2.43 To accelerate
the setting of high-alumina cement, lithium salts can be used.2.57
Because of the rapid setting just described, in construction it is essential to make
sure that the two cements do not accidentally come in contact with one another. Thus,
placing concrete made with one type of cement against concrete made with the other
must be delayed by at least 24 hours if high-alumina cement was cast first, or 3 to 7
days if the earlier concrete was made with Portland cement. Contamination through
plant or tools must also be avoided.
It may be noted that, for equal mix proportions, high-alumina cement produces a
somewhat more workable mix than when Portland cement is used. This may be due to

101
the lower total surface area of high-alumina cement particles, which have a ‘smoother’
surface than Portland cement particles, because high-alumina cement is produced by
complete fusion of the raw materials. On the other hand, superplasticizers do not give
good mobility and also adversely affect strength.2.74
Creep of high-alumina cement concrete has been found to differ little from the
creep of Portland cement concretes when the two are compared on the basis of the
stress/strength ratio.2.22
Composition and hydration
The main cementitious compounds are calcium aluminates of low basicity primarily
CA and also C12A7.2.32 Other compounds are also present: C6A4.FeO.S and an
isomorphous C6A4.MgO.S.2.13 The amount of C2S or C2AS does not account for more
than a few per cent, and there are, of course, minor compounds present, but no free
lime can exist. Thus unsoundness is never a problem in high-alumina cement although
BS 915 : 1972 (1983) prescribes the conventional Le Chatelier test.
The hydration of CA, which has the highest rate of strength development, results in
the formation of CAH10, a small quantity of C2AH8, and of alumina gel (Al2O3.aq).
With time, these hexagonal CAH10 crystals, which are unstable both at normal and at
higher temperatures, become transformed into cubic crystals of C3AH6 and alumina
gel. This transformation is encouraged by a higher temperature and a higher
concentration of lime or a rise in alkalinity.2.14
C12A7, which also hydrates rapidly, is believed to hydrate to C2AH8. The compound
C2S forms C-S-H, the lime liberated by hydrolysis reacting with excess alumina; no
Ca(OH)2 exists. The reactions of hydration of the other compounds, particularly those
containing iron, have not been determined with any degree of certainty, but the iron
held in glass is known to be inert.2.15 Iron compounds are useful as a flux in the
manufacture of high-alumina cement.
The water of hydration of high-alumina cement is calculated to be up to 50 per
cent of the mass of the dry cement,2.6 which is about twice as much as the water
required for the hydration of Portland cement, but mixes with a water/cement ratio as
low as 0.35 are practicable and indeed desirable. The pH of pore solution in high-
alumina cement paste is between 11.4 and 12.5.2.8
Conversion of high-alumina cement
The high strength of high-alumina cement concrete referred to on p. 93 is reached
when the hydration of CA results in the formation of CAH10 with a small quantity of
C2AH8 and of alumina gel (Al2O3.aq). The hydrate CAH10 is, however, chemically
unstable both at higher and normal temperatures and becomes transformed into
C3AH6 and alumina gel. This change is known as conversion, and, because the
symmetry of the crystal systems is pseudo-hexagonal for the decahydrate and cubic
for the sesquihydrate, one can refer to it as the change from the hexagonal to cubic
form.
An important feature of hydration of high-alumina cement is that, at higher
temperatures, only the cubic form of the calcium aluminate hydrate can exist; at room
temperature, either form can exist, but the hexagonal crystals spontaneously, albeit
slowly, convert to the cubic form. Because they undergo a spontaneous change, the
hexagonal crystals can be said to be unstable at room temperature, the final product of
the reactions of hydration being the cubic form. Higher temperature speeds up the

102
process; when the periods of exposure to a higher temperature are intermittent, their
effect is cumulative.2.18 This then is conversion: an unavoidable change of one form of
calcium aluminate hydrate to another, and it is only reasonable to add that this type of
change is not an uncommon phenomenon in nature.
Before discussing the significance of conversion, we should briefly describe the
reaction. Conversion both of CAH10 and of C2AH8 proceeds direct; for instance:

3CAH10 → C3AH6 + 2AH3 + 18H.

It should be noted that, although water appears as a product of the reaction,

conversion can take place only in the presence of water and not in desiccated concrete
because redissolving and reprecipitation are involved. As far as neat cement paste is
concerned, it has been found2.46 that, in sections thicker than 25 mm, the interior of the
hydrating cement has an equivalent relative humidity of 100 per cent regardless of the
environmental humidity, so that conversion can take place. The influence of the
ambient humidity is thus only on concrete near the surface.
The cubic product of conversion, C3AH6, is stable in a solution of calcium
hydroxide at 25 °C but reacts with a mixed Ca(OH)2–CaSO4 solution to form
3CaO.Al2O3.3CaSO4.31H2O both at 25 °C and at higher temperatures.2.47
The degree of conversion is estimated from the percentage of C3AH6 present as a
proportion of the sum of the cubic and hexagonal hydrates taken together, i.e. the
degree of conversion (per cent) is

The relative masses of the compounds are derived from the measurements of
endothermic peaks in a differential-thermal analysis thermogram.
However, unless the determination can be made under CO2-free conditions, there is
a risk of decomposition of C3AH6 into AH3. The degree of conversion can be
determined also in terms of the latter compound because, fortuitously, the masses of
C3AH6 and AH3 produced in conversion are not very different. Thus we can write: the
degree of conversion (per cent) is

While the two expressions do not give exactly the same result, at high degrees of
conversion the difference is not significant. Most laboratories report the result to the
nearest 5 per cent. Concrete which has converted about 85 per cent would be
considered as fully converted.
The rate of conversion depends on temperature; some actual data are shown
in Table 2.9. The relation2.46 between the time necessary for one-half of the CAH10 to
convert and the temperature of storage of 13 mm ( in.) cubes of neat cement paste
with a water/cement ratio of 0.26 is shown in Fig. 2.5. It is likely that, for the more
porous concretes of practical mix proportions, the periods are much shorter as full
conversion has been observed after some 20 years at 20 °C or thereabouts. Thus data
on neat cement pastes with very low water/cement ratios should be used
circumspectly, but they are nevertheless of scientific interest.

103
 Table 2.9. Development of Conversion with Age2.51 (Crown copyright)

Fig. 2.5. Time for half-conversion of neat high-alumina cement pastes cured at
various temperatures (13 mm ( in.) cubes)2.46 (Crown copyright)

104
 The practical interest in conversion lies in the fact that it leads to a loss of strength
of high-alumina cement concrete. The explanation of this is in terms of the
densification of the calcium aluminate hydrates: typically, the density would be 1.72
g/ml for CAH10 and 2.53 for C3AH6. Thus, under conditions such that the overall
dimensions of the body are constant (as is the case in set cement paste), conversion,
with the concomitant internal release of water, results in an increase in the porosity of
the paste. Numerous proofs of this are available, a particularly convincing one being
the measurement of air permeability of converted compared with unconverted high-
alumina cement concrete2.48 (see Fig. 2.6).

Fig. 2.6. Air flow through concrete: (a) unconverted high-alumina cement
concrete; (b) converted high-alumina cement concrete; (c) Portland cement
concrete (temperature 22 to 24 °C (72 to 75 °F), relative humidity 36 to 41 per
cent; pressure difference 10.7 kPa)2.48
As shown on p. 279, the strength of hydrated cement paste or of concrete is very
strongly affected by its porosity; porosity of 5 per cent can reduce the strength by
more than 30 per cent, and a 50 per cent reduction in strength would be caused by a
porosity of about 8 per cent. This magnitude of porosity of concrete can be induced by
conversion in high-alumina cement concrete.
It follows that, because conversion takes place in concretes and mortars of any mix
proportions, they lose strength when exposed to a higher temperature, and the general
pattern of the strength loss versus time is similar in all cases. However, the degree of
loss is a function of the water/cement ratio of the mix, as shown in Fig. 2.7. The mix
proportions and percentage loss are given in Table 2.10. It is clear that the loss, either
in megapascals (or psi) or as a fraction of the strength of cold-cured concrete, is
smaller in mixes with low water/cement ratios than in mixes with high water/cement
ratios.2.33

105
 Fig. 2.7. Influence of the water/cement ratio on the strength of high-alumina
cement concrete cubes cured in water at 18 and 40 °C for 100 days

Table 2.10. Influence of Water/Cement Ratio on Loss of Strength on Conversion

106
 It may be observed that the shape of the strength versus water/cement ratio curves
for storage at 18 °C (Fig. 2.7) is dissimilar from the usual curves for Portland cement
concretes. This is characteristic of concretes made with high-alumina cement, and has
been confirmed also for cylinders both of standard size2.17 and other height/diameter
ratios.2.22
The values shown in Fig. 2.7 are no more than typical, and clearly some variation
would be found with different cements, but the pattern of behaviour is the same in all
cases. It is important to note that the residual strength of mixes with moderate and
high water/cement ratios, say over 0.5, may be so low as to be unacceptable for most
structural purposes.
A brief historical note on the structural use of high-alumina cement may be in
order. Because of the very high early strength of concrete made with high-alumina
cement, it was used in the manufacture of prestressed concrete units. Neville’s
warnings2.33 about the dangers consequent upon conversion were ignored, but they
were shown to be true. Structural failures occurred in England in the early 1970s, and
consequently, all structural use of high-alumina cement was withdrawn from British
codes. In most other countries, too, high-alumina cement is not used in structural
concrete. Nevertheless, failures of old high-alumina cement concrete occurred in
Spain in the early 1990s. The European standard BS EN 14647-2006 deals with HAC
but it contains an annex giving advice on use of HAC; in my opinion, such advice on
structural use of HAC from a body writing a material specification is outside its remit.
Arguments to the effect that, at a water/cement ratio not exceeding 0.40 and a
cement content of not less than 400 kg/m3 (680 lb/yd3), the strength after conversion is
still adequate are not convincing. To begin with, under practical conditions of
manufacture of concrete, it is not possible to guarantee that the specified
water/cement ratio will not be occasionally exceeded by 0.05 or even by 0.10; this has
been repeatedly demonstrated2.49 (see also p. 744). It should be noted that the strength
of converted high-alumina cement is more sensitive to changes in the water/cement
ratio than before conversion; this is illustrated in Fig. 2.8 based on the data of
George.2.50

107
 Fig. 2.8. Influence of the water/cement ratio upon the strength of high-alumina
cement concrete, before and after conversion, relative to the strength after
conversion of concrete with a water/cement ratio of 0.4 (based on ref. 2.50)
Under certain moisture conditions, following conversion, hydration of the hitherto
unhydrated cement leads to some increase in strength. However, the conversion of the
newly formed hexagonal hydrates leads to a renewed and continuing loss of strength.
Thus, the strength drops below the 24-hour value. This occurs at the age of 8 to 10
years in concrete with a water/cement ratio of 0.4, and even later if the water/cement
ratio is lower.2.78 In any case, from the structural point of view, it is the lowest strength
at any time in the life of the concrete that is critical.
The loss of strength is lower under dry conditions, but in concrete of substantial
thickness the conditions are not dry. An indirect proof that within a large mass of rich
concrete there is present adequate water for chemical reaction is afforded by
Hobbs2.75 who found that, in concrete with a Portland cement content of 500 to 550
kg/m3, kept sealed, there was enough water available for expansive alkali–silica
reaction to take place. Collins and Gutt2.78 reported that wet, or perhaps even
occasionally wet, concrete may have a strength 10 to 15 MPa lower than dry concrete.
Occasional wetting, by accident or, for instance, applied to extinguish a fire may
occur in almost every building.

108
 These results of Collins and Gutt2.78 were the outcome of an investigation at the
Building Research Establishment, started in 1964, which confirmed in essence the
statements and extrapolations made in 1963 by Neville.2.33 Menzies2.84 described the
advice given in earlier codes of practice as an error. The saga of problems with use of
HAC is given in ref.2.85
The second argument concerning the structural use of converted high-alumina
cement concrete, even if it is of adequate strength, is that converted high-alumina
cement paste is more porous and therefore more liable to chemical attack than before
conversion. This applies in particular to sulfate attack. If the sulfate ions penetrate
through the outer protective skin of high-alumina cement concrete (associated with
drying out), expansive reaction with C3AH6 takes place;2.79 it is only unconverted
CAH10 that is inert with respect to sulfates.
Moreover, chemical attack may then produce a further loss in strength,2.81 but the
chemical reactions involved necessitate the presence of water. As mentioned on p. 93,
percolating water can bring with it sodium or potassium hydroxide, which accelerate
conversion, in addition to decomposition of the products of hydration. If carbon
dioxide is also present, calcium carbonate is formed, and the alkali hydroxide is
regenerated to attack further the hydrated cement paste.2.82 Under some circumstances,
complete decomposition of calcium aluminate hydrates can result. The reactions are
written as follows:2.83

K2CO3 + CaO.Al2O3.aq → CaCO3 + K2O.Al2O3

CO2 + K2O.Al2O3 + aq → K2CO3 + Al2O3.3H2O.

Thus, because the alkalis are only a carrier, the overall reaction can be written as:

CO2 + CaO.Al2O3.aq → CaCO3 + Al2O3.3H2O.

It can be said, therefore, that high-alumina cement undergoes carbonation, but its
nature is not the same as that of Portland cement (see p. 499).
British Standards do not allow structural use of high-alumina cement. In the United
States, the Strategic Highway Research Program2.37 decided not to consider high-
alumina cement concrete because of the consequences of conversion. The cement has,
however, specialized applications. One of these is in mines for roof support. Here, a
two-slurry system containing high-alumina cement, calcium sulfate, lime and
appropriate admixtures leads to the development of ettringite, which has a substantial
early strength:2.72

3CA + 3C H2 + 2C + 26H → C6A 3H32.

denotes SO3

Madjumdar et al.2.73 developed a blended high-alumina and ground granulated

blastfurnace slag cement (in equal proportions by mass) in an attempt to avoid
conversion problems. The slag removes lime from the solution so that the formation
of C3AH6 is hindered and C2ASH8 is the main hydrate formed in the longer term.
However, such a blended cement does not develop the very high early strength which
is characteristic of high-alumina cement – its apparent forte; this may be the reason
why this blended cement has not been commercially produced.

109
Refractory properties of high-alumina cement
High-alumina cement concrete is one of the foremost refractory materials, but it is
important to be clear about its performance over the full temperature range. Between
room temperature and about 500 °C, high-alumina cement concrete loses strength to a
greater extent than Portland cement concrete; then up to 800 °C the two are
comparable; but above about 1000 °C high-alumina cement gives excellent
performance. Figure 2.9 shows the behaviour of high-alumina cement concrete made
with four different aggregates at temperatures up to 1100 °C.2.52 The minimum strength
varies between 5 and 26 per cent of the original value but, depending on the type of
aggregate, above 700 to 1000 °C, there is a gain in strength due to the development of
the ceramic bond. This bond is established by solid reactions between the cement and
the fine aggregate, and increases with an increase in temperature and with the
progress of the reactions.

Fig. 2.9. Strength of high-alumina cement concretes made with different

aggregates as a function of temperature2.52
As a result, high-alumina cement concrete can withstand very high temperatures:
with crushed firebrick aggregate up to about 1350 °C, and with special aggregates,
such as fused alumina or carborundum, up to 1600 °C. A temperature as high as
1800 °C can be withstood over prolonged periods of time by concrete made from
special white calcium aluminate cement with fused alumina aggregate. This cement is
made using alumina as a raw material and contains 70 to 80 per cent of A12O3, 20 to
25 per cent of lime, and only about 1 per cent of iron and silica; the composition of
the cement approaches C3A5. It is appropriate to mention that the price of such cement
is very high.

110
 Refractory concrete made with high-alumina cement has a good resistance to acid
attack (e.g. acids in flue gases), the chemical resistance being in fact increased by
firing at 900 to 1000 °C.2.16 The concrete can be brought up to service temperature as
soon as it has hardened, that is, it does not have to be pre-fired. While refractory
brickwork expands on heating and, therefore, needs expansion joints, high-alumina
cement concrete can be cast monolithically, or with butt joints only (at 1 to 2 m), to
exactly the required shape and size. The reason for this is that the loss of water on first
firing results in a contraction approximately equal to the thermal expansion on heating,
so that the net dimensional change (depending on aggregate) is small. Upon
subsequent cooling, for instance, during the shut-down of a plant, butt joints would
open slightly due to the thermal contraction but they would close up again on re-
heating. It is worth noting that refractory high-alumina cement concrete can withstand
a considerable thermal shock. Refractory linings can be made by shotcreting high-
alumina cement mortar.
For insulating purposes, when temperatures up to about 950 °C are expected,
lightweight concrete can be made with high-alumina cement and lightweight
aggregate. Such concrete has a density of 500 to 1000 kg/m3 (30 to 60 lb/ft3) and a
thermal conductivity of 0.21 to 0.29 J/m2s °C/m (0.12 to 0.17 Btu/ft2h °F/ft).
References
2.1. U.S. BUREAU OF RECLAMATION, Concrete Manual, 8th Edn (Denver, Colorado,
1975).
2.2. H. WOODS, Rational development of cement specifications, J. Portl. Cem.
Assoc. Research and Development Laboratories, 1, No. 1, pp. 4–11 (Jan.
1959).
2.3. W. H. PRICE, Factors influencing concrete strength, J. Amer. Concr. Inst., 47,
pp. 417–32 (Feb. 1951).
2.4. H. F. GONNERMAN and W. LERCH, Changes in characteristics of portland
cement as exhibited by laboratory tests over the period 1904 to 1950, ASTM
Sp. Publ. No. 127 (1951).
2.5. A. NEVILLE, Cementitious materials – a different viewpoint, Concrete
International, 16, No. 7, pp. 32–3 (1994).
2.6. F. M. LEA, The Chemistry of Cement and Concrete (London, Arnold, 1970).
2.7. R. H. BOGUE, Portland cement, Portl. Cem. Assoc. Fellowship Paper No. 53,
pp. 411–31 (Washington DC, August 1949).
2.8. S. GOÑI, G. ANDRADE and C. L. PAGE, Corrosion behaviour of steel in high
alumina cement mortar samples: effect of chloride, Cement and Concrete
Research, 21, No. 4, pp. 635–46 (1991).
2.9. ACI 225R-91, Guide to the selection and use of hydraulic cements, ACI
Manual of Concrete Practice, Part 1: Materials and General Properties of
Concrete, 29 pp. (Detroit, Michigan, 1994).
2.10. R. W. NURSE, Hydrophobic cement, Cement and Lime Manufacture, 26, No.
4, pp. 47–51 (London, July 1953).
2.11. U. W. STOLL, Hydrophobic cement, ASTM Sp. Tech. Publ., No. 205, pp. 7–15
(1958).

111
 2.12. J. BENSTED, An investigation of the early hydration characteristics of some
low alkali Portland cements, Il Cemento, 79, No. 3, pp. 151–8 (July 1992).
2.13. T. W. PARKER, La recherche sur la chimie des ciments au Royaume-Uni
pendant les années d’après-guerre, Revue Génerale des Sciences
Appliquées, 1, No. 3, pp. 74–83 (1952).
2.14. H. LAFUMA, Quelques aspects de la physicochimie des ciments
alumineux, Revue Génerale des Sciences Appliquées, 1, No. 3, pp. 66–74.
2.15. F. M. LEA, Cement and Concrete, Lecture delivered before the Royal
Institute of Chemistry, London, 19 Dec. 1944 (Cambridge, W. Heffer and
Sons, 1944).
2.16. A. V. HUSSEY and T. D. ROBSON, High-alumina cement as a constructional
material in the chemical industry, Symposium on Materials of Construction
in the Chemical Industry, Birmingham, Soc. Chem. Ind., 1950.
2.17. A. M. NEVILLE, Tests on the strength of high-alumina cement concrete, J.
New Zealand Inst. E., 14, No. 3, pp. 73–6 (March 1959).
2.18. A. M. NEVILLE, The effect of warm storage conditions on the strength of
concrete made with high-alumina cement, Proc. Inst. Civ. Engrs., 10, pp.
185–92 (London, June 1958).
2.19. E. W. BENNETT and B. C. COLLINGS, High early strength concrete by means of
very fine Portland cement, Proc. Inst. Civ. Engrs, pp. 1–10 (July 1969).
2.20. H. UCHIKAWA, S. UCHIDA, K. OGAWA and S. HANEHARA, Influence of the amount,
state and distribution of minor constituents in clinker on the color of white
cement, Il Cemento, 3, pp. 153–68 (1986).
2.21. AMERICAN PETROLEUM INSTITUTE Specification of Oil-Well Cements and Cement
Additives, 20 pp. (Dallas, Texas, 1992).
2.22. A. M. NEVILLE and H. KENINGTON, Creep of aluminous cement concrete, Proc.
4th Int. Symp. on the Chemistry of Cement, pp. 703–8 (Washington DC,
1960).
2.23. K. KOHNO and K. ARAKI, Fundamental properties of stiff consistency concrete
made with jet cement, Bulletin of Faculty of Engineering, Tokushima
University, 10, Nos 1 and 2, pp. 25–36 (1974).
2.24. RILEM DRAFT RECOMMENDATIONS TC FAB-67, Test methods for determining the
properties of fly ash and of fly ash for use in building materials, Materials
and Structures, 22, No. 130, pp. 299–308 (1989).
2.25. F. SYBERTZ, Comparison of different methods for testing the pozzolanic
activity of fly ashes, in Fly Ash, Silica Fume, Slag, and Natural Pozzolans in
Concrete, Vol. 1, Ed. V. M. Malhotra, ACI SP-114, pp. 477–97 (Detroit,
Michigan, 1989).
2.26. P. K. MEHTA, Rice husk ash – a unique supplementary cementing material,
in Advances in Concrete Technology, Ed. V. M. Malhotra, Energy, Mines
and Resources, MSL 92-6(R) pp. 407–31 (Ottawa, Canada, 1992).
2.27. D. M. ROY, Hydration of blended cements containing slag, fly ash, or silica
fume, Proc. of Meeting Institute of Concrete Technology, Coventry, 29 pp.
(29 April–1 May 1987).

112
 2.28. F. MAZLUM and M. UYAN, Strength of mortar made with cement containing
rice husk ash and cured in sodium sulphate solution, in Fly Ash, Silica Fume,
Slag, and Natural Pozzolans in Concrete, Vol. 1, Ed. V. M. Malhotra, ACI
SP-132, pp. 513–31 (Detroit, Michigan, 1993).
2.29. P. DUTRON, Present situation of cement standardization in Europe, Blended
Cements, Ed. G. Frohnsdorff, ASTM Sp. Tech. Publ. No. 897, pp. 144–53
(Philadelphia, 1986).
2.30. G. E. MONFORE and G. J. VERBECK, Corrosion of prestressed wire in concrete, J.
Amer. Concr. Inst. 57, pp. 491–515 (Nov. 1960).
2.31. E. BURKE, Discussion on comparison of chemical resistance of
supersulphated and special purpose cements, Proc. 4th Int. Symp. on the
Chemistry of Cement, pp. 877–9 (Washington DC, 1960).
2.32. P. LHOPITALLIER, Calcium aluminates and high-alumina cement, Proc. 4th Int.
Symp. on the Chemistry of Cement, pp. 1007–33 (Washington DC, 1960).
2.33. A. M. NEVILLE, A study of deterioration of structural concrete made with
high-alumina cement, Proc. Inst. Civ. Engrs, 25, pp. 287–324 (London, July
1963).
2.34. U.S. BUREAU OF RECLAMATION, Concrete Manual, 5th Edn (Denver, Colorado,
1949).
2.35. AGRÉMENT BOARD, Certificate No. 73/170 for Swiftcrete ultra high early
strength cement (18 May 1973).
2.36. E. W. BENNETT and D. R. LOAT, Shrinkage and creep of concrete as affected
by the fineness of Portland cement, Mag. Concr. Res., 22, No. 71, pp. 69–78
(1970).
2.37. STRATEGIC HIGHWAY RESEARCH PROGRAM, High Performance Concretes: A State-
of-the-Art Report, NRC, SHRP-C/FR-91-103, 233 pp. (Washington DC,
1991).
2.38. C. J. LYNSDALE and J. G. CABRERA, Coloured concrete: a state of the art
review, Concrete, 23, No. 1, pp. 29–34 (1989).
2.39. E. DOUGLAS and V. M. MALHOTRA, A Review of the Properties and Strength
Development of Non-ferrous Slags and Portland Cement Binders,
CANMET Report, No. 85-7E, 37 pp. (Canadian Govt Publishing Centre,
Ottawa, 1986).
2.40. S. M. BUSHNELL-WATSON and J. H. SHARP, On the cause of the anomalous
setting behaviour with respect to temperature of calcium aluminate
cements, Cement and Concrete Research, 20, No. 5, pp. 677–86 (1990).
2.41. E. DOUGLAS, A. BILODEAU and V. M. MALHOTRA, Properties and durability of
alkali-activated slag concrete, ACI Materials Journal, 89, No. 5, pp. 509–16
(1992).
2.42. D. W. QUINION, Superplasticizers in concrete – a review of international
experience of long-term reliability, CIRIA Report 62, 27 pp. (London,
Construction Industry Research and Information Assoc., Sept. 1976).
2.43. ACI 517.2R-87, (Revised 1992), Accelerated curing of concrete at
atmospheric pressure – state of the art, ACI Manual of Concrete Practice
Part 5 – 1992: Masonry, Precast Concrete, Special Processes, 17 pp.
(Detroit, Michigan, 1994).

113
 2.44. E. ZIELINSKA, The influence of calcium carbonate on the hydration process in
some Portland cement constituents (3Ca.Al2O3 and 4CaO.Al2O3.Fe2O3), Prace
Instytutu Technologii i Organizacji Produkcji Budowlanej, No. 3 (Warsaw
Technical University, 1972).
2.45. G. M. IDORN, The effect of slag cement in concrete, NRMCA Publ. No. 167,
10 pp. (Maryland, USA, April 1983).
2.46. H. G. MIDGLEY, The mineralogy of set high-alumina cement, Trans. Brit.
Ceramic Soc., 66, No. 4, pp. 161–87 (1967).
2.47. A. KELLEY, Solid–liquid reactions amongst the calcium aluminates and
sulphur aluminates, Canad. J. Chem., 38, pp. 1218–26 (1960).
2.48. H. MARTIN, A. RAUEN and P. SCHIESSL, Abnahme der Druckfestigkeit von
Beton aus Tonerdeschmelzzement, Aus Unseren Forschungsarbeiten, III,
pp. 34–7 (Technische Universität München, Inst. für Massivbau, Dec. 1973).
2.49. A. M. NEVILLE in collaboration with P. J. Wainwright, High-alumina Cement
Concrete, 201 pp. (Lancaster, Construction Press, Longman Group, 1975).
2.50. C. M. GEORGE, Manufacture and performance of aluminous cement: a new
perspective, Calcium Aluminate Cements, Ed. R. J. Mangabhai, Proc. Int.
Symp., Queen Mary and Westfield College, University of London, pp. 181–
207 (London, Chapman and Hall, 1990).
2.51. D. C. TEYCHENNÉ, Long-term research into the characteristics of high-alumina
cement concretes, Mag. Conor. Res., 27, No. 91, pp. 78–102 (1975).
2.52. N. G. ZOLDNERS and V. M. MALHOTRA, Discussion of reference 2.33, Proc. Inst.
Civ. Engrs, 28, pp. 72–3 (May 1964).
2.53. J. AMBROISE, S. MARTIN-CALLE and J. PÉRA, Pozzolanic behaviour of thermally
activated kaolin, in Fly Ash, Silica Fume, Slag, and Natural Pozzolans in
Concrete, Vol. 1, Ed. V. M. Malhotra, ACI SP-132, pp. 731–48 (Detroit,
Michigan, 1993).
2.54. W. H. DUDA, Cement-Data-Book, 2, 456 pp. (Berlin, Verlag GmbH, 1984).
2.55. K. W. NASSER and H. M. MARZOUK, Properties of mass concrete containing fly
ash at high temperatures, J. Amer. Concr. Inst., 76, No. 4, pp. 537–50 (April
1979).
2.56. ACI 3R-87, Ground granulated blast-furnace slag as a cementitious
constituent in concrete, ACI Manual of Concrete Practice, Part 1: Materials
and General Properties of Concrete, 16 pp. (Detroit, Michigan, 1994).
2.57. T. NOVINSON and J. CRAHAN, Lithium salts as set accelerators for refractory
concretes: correlation of chemical properties with setting times, ACI
Materials Journal, 85, No. 1, pp. 12–16 (1988).
2.58. J. DAUBE and R. BARKER, Portland blast-furnace slag cement: a
review, Blended Cements, Ed. G. Frohnsdorff, ASTM Sp. Tech. Publ. No.
897, pp. 5–14 (Philadelphia, 1986).
2.59. G. FRIGIONE, Manufacture and characteristics of Portland blast-furnace slag
cements, Blended Cements, Ed. G. Frohnsdorff, ASTM Sp. Tech. Publ. No.
897, pp. 15–28 (Philadelphia, 1986).

114
 2.60. M. N. A. SAAD, W. P. DE ANDRADE and V. A. PAULON, Properties of mass
concrete containing an active pozzolan made from clay, Concrete
International, 4, No. 7, pp. 59–65 (1982).
2.61. K. KOHNO et al., Mix proportion and compressive strength of concrete
containing extremely finely ground silica, Cement Association of Japan, No.
44, pp. 157–80 (1990).
2.62. B. P. HUGHES, PFA fineness and its use in concrete, Mag. Concr. Res., 41, No.
147, pp. 99–105 (1989).
2.63. W. H. PRICE, Pozzolans – a review, J. Amer. Concr. Inst., 72, No. 5, pp. 225–
32. (1975).
2.64. ACI 3R-87, Use of fly ash in concrete, ACI Manual of Concrete
Practice, Part 1: Materials and General Properties of Concrete, 29 pp.
(Detroit, Michigan, 1994).
2.65. Y. EFES and P. SCHUBERT, Mörtel- und Betonversuche mit einem
Schnellzement, Betonwerk und Fertigteil-Technik, No. 11, pp. 541–5 (1976).
2.66. P.-C. AÏTCIN, Ed., Condensed Silica Fume, Faculté de Sciences Appliquées,
Université de Sherbrooke, 52 pp. (Sherbrooke, Canada, 1983).
2.67. ACI COMMITTEE 226, Silica fume in concrete: Preliminary report, ACI
Materials Journal, 84, No. 2, pp. 158–66 (1987).
2.68. D. G. PARKER, Microsilica concrete, Part 2: in use, Concrete, 20, No. 3, pp.
19–21 (1986).
2.69. V. M. MALHOTRA, G. G. CARRETTE and V. SIVASUNDARAM, Role of silica fume in
concrete: a review, in Advances in Concrete Technology, Ed. V. M.
Malhotra, Energy, Mines and Resources, MSL 92-6(R) pp. 925–91 (Ottawa,
Canada, 1992).
2.70. M. D. COHEN, Silica fume in PCC: the effects of form on engineering
performance, Concrete International, 11, No. 11, pp. 43–7 (1989).
2.71. K. H. KHAYAT and P. C. AÏTCIN, Silica fume in concrete – an overview, in Fly
Ash, Silica Fume, Slag, and Natural Pozzolans in Concrete, Vol. 2, Ed. V.
M. Malhotra, ACI SP-132, pp. 835–72 (Detroit, Michigan, 1993).
2.72. S. A. BROOKS and J. H. SHARP, Ettringite-based cements, Calcium Aluminate
Cements, Ed. R. J. Mangabhai, Proc. Int. Symp., Queen Mary and Westfield
College, University of London, pp. 335–49 (London, Chapman and Hall,
1990).
2.73. A. J. MAJUMDAR, R. N. EDMONDS and B. SINGH, Hydration of calcium
aluminates in presence of granulated blastfurnace slag, Calcium Aluminate
Cements, Ed. R. J. Mangabhai, Proc. Int. Symp., Queen Mary and Westfield
College, University of London, pp. 259–71 (London, Chapman and Hall,
1990).
2.74. V. S. RAMACHANDRAN, Ed., Concrete Admixtures Handbook:
Properties, Science and Technology, 626 pp. (New Jersey, Noyes
Publications, 1984).
2.75. D. W. HOBBS, Alkali-Silica Reaction in Concrete, 183 pp. (London, Thomas
Telford, 1988).

115
 2.76. M. COLLEPARDI, G. BALDINI and M. PAURI, The effect of pozzolanas on the
tricalcium aluminate hydration, Cement and Concrete Research, 8, No. 6, pp.
741–51 (1978).
2.77. F. MASSAZZA and U. COSTA, Aspects of the pozzolanic activity and properties
of pozzolanic cements, Il Cemento, 76, No. 1, pp. 3–18 (1979).
2.78. R. J. COLLINS and W. GUTT, Research on long-term properties of high alumina
cement concrete, Mag. Concr. Res., 40, No. 145, pp. 195–208 (1988).
2.79. N. J. CRAMMOND, Long-term performance of high alumina cement in sulphate-
bearing environments, Calcium Aluminate Cements, Ed. R. J. Mangabhai,
Proc. Int. Symp., Queen Mary and Westfield College, University of London,
pp. 208–21 (London, Chapman and Hall, 1990).
2.80. D. J. COOK, R. P. PARMA and S. A. DAMER, The behaviour of concrete and
cement paste containing rice husk ash, Proc. of a Conference on Hydraulic
Cement Pastes: Their Structure and Properties, pp. 268–82 (London,
Cement and Concrete Assoc., 1976).
2.81. T. D. ROBSON, The characteristics and applications of mixtures of Portland
and high-alumina cements, Chemistry and Industry, No. 1, pp. 2–7 (London,
5 Jan. 1952).
2.82. BUILDING RESEARCH ESTABLISHMENT, Assessment of chemical attack of high
alumina cement concrete, Information Paper, IP 22/81, 4 pp. (Watford,
England, Nov. 1981).
2.83. F. M. LEA, Effect of temperature on high-alumina cement, Trans. Soc.
Chem., 59, pp. 18–21 (1940).
2.84. J. B. MENZIES, Hazards, risks and structural safety, Structural Engineer, 10,
No. 21, pp. 357–63 (1995).
2.85. A. M. NEVILLE, History of high-alumina cement, Engineering History and
Heritage, Inst. Civil Engineers, EH2, pp. 81–91 and 93–101 (2009).

116
Chapter 3. Properties of aggregate

Because at least three-quarters of the volume of concrete is occupied by aggregate, it

is not surprising that its quality is of considerable importance. Not only may the
aggregate limit the strength of concrete, as aggregate with undesirable properties
cannot produce strong concrete, but the properties of aggregate greatly affect the
durability and structural performance of concrete.
Aggregate was originally viewed as an inert material dispersed throughout the
cement paste largely for economic reasons. It is possible, however, to take an opposite
view and to look on aggregate as a building material connected into a cohesive whole
by means of the cement paste, in a manner similar to masonry construction. In fact,
aggregate is not truly inert and its physical, thermal, and sometimes also chemical
properties influence the performance of concrete.
Aggregate is cheaper than cement and it is, therefore, economical to put into the
mix as much of the former and as little of the latter as possible. But economy is not
the only reason for using aggregate: it confers considerable technical advantages on
concrete, which has a higher volume stability and better durability than hydrated
cement paste alone.
General classification of aggregates
The size of aggregate used in concrete ranges from tens of millimetres down to
particles less than one-tenth of a millimetre in cross-section. The maximum size
actually used varies but, in any mix, particles of different sizes are incorporated, the
particle size distribution being referred to as grading. In making low-grade concrete,
aggregate from deposits containing a whole range of sizes, from the largest to the
smallest, is sometimes used; this is referred to as all-in or pit-run aggregate. The
alternative, always used in the manufacture of good quality concrete, is to obtain the
aggregate in at least two size groups, the main division being between fine aggregate,
often called sand (for example, in BS EN 12620 : 2002), not larger than 4 mm
or in., and coarse aggregate, which comprises material at least 5 mm or in. in
size. In the United States, the division is made at No. 4 ASTM sieve, which is 4.75
mm ( in.) in size (see Table 3.14). More will be said about grading later, but this
basic division makes it possible to distinguish in the ensuing description between fine
and coarse aggregate. It should be noted that the use of the term aggregate (to mean
coarse aggregate) in contradistinction to sand is not correct.
Natural sand is generally considered to have a lower size limit of 70 or 60 μm.
Material between 60 μm and 2 μm is classified as silt, and particles smaller still are
termed clay. Loam is a soft deposit consisting of sand, silt, and clay in about equal
proportions. Although the content of particles smaller than 75 μm is usually reported
globally, the influence of silt and of clay on the properties of the resultant concrete is
often significantly different not only because these particles differ in size but also in
composition. Methods of determining the proportion of material smaller than 75 μm
and 20 μm, respectively, are prescribed in BS 812 : 103.1 : 1985 (2000) and BS 812 :
103.2 (2000).
All natural aggregate particles originally formed a part of a larger parent mass.
This may have been fragmented by natural processes of weathering and abrasion or
artificially by crushing. Thus, many properties of the aggregate depend entirely on the

117
properties of the parent rock, e.g. chemical and mineral composition, petrological
character, specific gravity, hardness, strength, physical and chemical stability, pore
structure, and colour. On the other hand, there are some properties possessed by the
aggregate but absent in the parent rock: particle shape and size, surface texture, and
absorption. All these properties may have a considerable influence on the quality of
the concrete, either fresh or in the hardened state.
It is only reasonable to add, however, that, although these different properties of
aggregate per se can be examined, it is difficult to define a good aggregate other than
by saying that it is an aggregate from which good concrete (for the given conditions)
can be made. While aggregate whose properties all appear satisfactory will always
make good concrete, the converse is not necessarily true and this is why the criterion
of performance in concrete has to be used. In particular, it has been found that
aggregate may appear to be unsatisfactory on some count but no trouble need be
experienced when it is used in concrete. For instance, a rock sample may disrupt on
freezing but need not do so when embedded in concrete, especially when the
aggregate particles are well covered by a hydrated cement paste of low permeability.
However, aggregate considered poor in more than one respect is unlikely to make a
satisfactory concrete, so that tests on aggregate alone are of help in assessing its
suitability for use in concrete.
Classification of natural aggregates
So far, we have considered only aggregate formed from naturally occurring materials,
and the present chapter deals almost exclusively with this type of aggregate.
Aggregate can, however, also be manufactured from industrial products: because
these artificial aggregates are generally either heavier or lighter than ordinary
aggregate they are considered in Chapter 13. Aggregates made from waste are
referred to on p. 696.
A further distinction can be made between aggregate reduced to its present size by
natural agents and crushed aggregate obtained by a deliberate fragmentation of rock.
From the petrological standpoint, the aggregates, whether crushed or naturally
reduced in size, can be divided into several groups of rocks having common
characteristics. The classification of BS 812 : 1 : 1975 is most convenient and is given
in Table 3.1. The group classification does not imply suitability of any aggregate for
concrete-making: unsuitable material can be found in any group, although some
groups tend to have a better record than others. It should also be remembered that
many trade and customary names of aggregates are in use, and these often do not
correspond to the correct petrographic classification. Rock types commonly used for
aggregates are listed in BS 812 : 102 : 1989, and BS 812 : 104 : 1994 (2000) covers
the methods of petrographic examination. BS 812 has been replaced by BS EN 932
and 933.

Table 3.1. Classification of Natural Aggregates According to Rock Type (BS 812 :
1 : 1975)

118
 ASTM Standard C 294-05 gives a description of some of the more common or
important minerals found in aggregates. Mineralogical classification is of help in
recognizing properties of aggregate but cannot provide a basis for predicting its
performance in concrete as there are no minerals universally desirable and few
invariably undesirable ones. The ASTM classification of minerals is summarized
below:
Silica minerals (quartz, opal, chalcedony, tridymite, cristobalite)
Feldspars

119
 Ferromagnesian minerals
Micaceous minerals
Clay minerals
Zeolites
Carbonate minerals
Sulfate minerals
Iron sulfide minerals
Iron oxides
The details of petrological and mineralogical methods are outside the scope of this
book, but it is important to realize that geological examination of aggregate is a useful
aid in assessing its quality and, in particular, in comparing a new aggregate with one
for which service records are available. Furthermore, adverse properties, such as the
presence of some unstable forms of silica, can be detected. Even small amounts of
minerals or of rocks may have a large influence on the quality of aggregate. In the
case of artificial aggregates, the influence of manufacturing methods and of
processing can also be studied. Detailed information on aggregate for concrete can be
found in ref. 3.38.
Sampling
Tests of various properties of aggregate are perforce performed on samples of the
material and, therefore, the results of the tests apply, strictly speaking, only to the
aggregate in the sample. Since, however, we are interested in the bulk of the
aggregate as supplied, or as available for supply, we should ensure that the sample is
typical of the average properties of the aggregate. Such a sample is said to be
representative and, to obtain it, certain precautions in procuring the sample have to be
observed.
No detailed procedures can, however, be laid down because the conditions and
situations involved in taking samples in the field can vary widely from case to case.
Nevertheless, an intelligent experimenter can obtain reliable results if he or she bears
in mind at all times that the sample taken is to be representative of the bulk of the
material considered. An instance of such care would be to use a scoop rather than a
shovel so as to prevent rolling off of particles of some sizes when the shovel is lifted.
The main sample is made up of a number of portions drawn from different parts of
the whole. The minimum number of these portions, called increments, is ten, and they
should add up to a mass not less than that given in Table 3.2 for particles of different
sizes, as prescribed by BS 812 : 102 : 1989 (replaced by BS EN 932-1 : 1997). If,
however, the source from which the sample is being obtained is variable or segregated,
a larger number of increments should be taken and a larger sample ought to be
dispatched for testing. This is particularly the case in stockpiles when increments have
to be taken from all parts of the pile, not only below its surface but also from the
centre of the pile.

Table 3.2. Minimum Mass of Samples for Testing (BS 812 : 102 : 1989)

120
 It is clear from Table 3.2 that the main sample can be rather large, particularly
when large-size aggregate is used, and so the sample has to be reduced before testing.
At all stages of reduction, it is necessary to ensure that the representative character of
the sample is retained so that the actual test sample has the same properties as the
main sample and ipso facto as the bulk of the aggregate.
There are two ways of reducing the size of a sample, each essentially dividing it
into two similar parts: quartering and riffling. For quartering, the main sample is
thoroughly mixed and, in the case of fine aggregate, dampened in order to avoid
segregation. The material is heaped into a cone and then turned over to form a new
cone. This is repeated twice, the material always being deposited at the apex of the
cone so that the fall of particles is evenly distributed round the circumference. The
final cone is flattened and divided into quarters. One pair of diagonally opposite
quarters is discarded, and the remainder forms the sample for testing or, if still too
large, can be reduced by further quartering. Care must be taken to include all fine
material in the appropriate quarter.
As an alternative, the sample can be split into halves using a riffler (Fig. 3.1). This
is a box with a number of parallel vertical divisions, alternate ones discharging to the
left and to the right. The sample is discharged into the riffler over its full width, and
the two halves are collected in two boxes at the bottom of the chutes on each side.
One half is discarded, and riffling of the other half is repeated until the sample is
reduced to the desired size. BS EN 12420 : 2000 describes a typical riffler. Riffling
gives less variable results than quartering.

121
 Fig. 3.1. Riffler
Particle shape and texture
In addition to the petrological character of aggregate, its external characteristics are of
importance, in particular the particle shape and surface texture. The shape of three-
dimensional bodies is rather difficult to describe, and it is, therefore, convenient to
define certain geometrical characteristics of such bodies.
Roundness measures the relative sharpness or angularity of the edges and corners
of a particle. Roundness is controlled largely by the strength and abrasion resistance
of the parent rock and by the amount of wear to which the particle has been subjected.
In the case of crushed aggregate, the particle shape depends not only on the nature of
the parent material but also on the type of crusher and its reduction ratio, i.e. the ratio
of the size of material fed into the crusher to the size of the finished product. A
convenient broad classification of roundness is that of BS 812 : 1 : 1975, given
in Table 3.3. The relevant current standard is BS EN 933-4 : 2008.

Table 3.3. Particle Shape Classification of BS 812-1 : 1975* with Examples

122
 *
Replaced by BS EN 933-3 : 1997
A classification sometimes used in the United States is as follows:

Because the degree of packing of particles, all of one size, depends on their shape,
the angularity of aggregate can be estimated from the proportion of voids in a sample
compacted in a prescribed way. British Standard BS 812-1 : 1995 defines the concept
of angularity number; this can be taken as 67 minus the percentage of solid volume in
a vessel filled with aggregate in a standard manner. The size of particles used in the
test must be controlled within narrow limits.
The number 67 in the expression for the angularity number represents the
percentage solid volume of the most rounded gravel, so that the angularity number
measures the percentage of voids in excess of that in the rounded gravel (i.e. 33). The
higher the number the more angular the aggregate, the range for practical aggregate
being between 0 and 11. The test for angularity is rarely used.
A development in measurement of angularity of aggregate, both coarse and fine
but of single size, is an angularity factor defined as the ratio of the solid volume of
loose aggregate to the solid volume of glass spheres of specified grading;3.41 thus, no
packing is involved and the attendant error is avoided. Various other indirect methods
of determination of the shape of fine aggregate have been critically reviewed by
Gaynor and Meininger3.63 but no generally accepted method is available.

123
 The void content of aggregate can be calculated from the change in the volume of
air when a known decrease in pressure is applied; hence, the volume of air, i.e. the
volume of interstitial space, can be calculated.3.52
A simple proof of the dependence of the percentage of voids on the shape of
particles is obtained from Fig. 3.2, based on Shergold’s3.1 data. The sample consisted
of a mixture of two aggregates, one angular, the other rounded, in varying proportions,
and it can be seen how increasing the proportion of rounded particles decreases the
percentage of voids. The volume of voids influences the density of the concrete which
can be achieved.

Fig. 3.2. Influence of angularity of aggregate on voids ratio3.1 (Crown copyright)

Another aspect of the shape of coarse aggregate is its sphericity, defined as a
function of the ratio of the surface area of the particle to its volume. Sphericity is
related to the bedding and cleavage of the parent rock, and is also influenced by the
type of crushing equipment when the size of particles has been artificially reduced.
Particles with a high ratio of surface area to volume are of particular interest as they
increase the water demand for a given workability of the concrete mix.
That the shape of fine aggregate particles influences the mix properties is without
doubt, angular particles requiring more water for a given workability, but an objective
method of measuring and expressing shape is not yet available despite attempts using
measurement of the projected surface area and other geometrical approximations.
As far as coarse aggregate is concerned, equidimensional shape of particles is
preferred because particles which significantly depart from such a shape have a larger
surface area and pack in an anisotropic manner. Two types of particles which depart
from equidimensional shape are of interest: elongated and flaky. The latter type can

124
also affect adversely the durability of concrete because flaky particles tend to be
oriented in one plane, with bleeding water and air voids forming underneath.
The mass of flaky particles expressed as a percentage of the mass of the sample is
called the flakiness index. Elongation index is similarly defined. Some particles are
both flaky and elongated, and are, therefore, counted in both categories.
The classification is made by means of simple gauges described in BS 812-105.1 :
1989 and BS EN 933-3 : 1997. BS EN 12620 : 2002 uses different dimensional ratios.
The division is based on the rather arbitrary assumption that a particle is flaky if its
thickness (least dimension) is less than 0.6 times the mean sieve size of the size
fraction to which the particle belongs. Similarly, a particle whose length (largest
dimension) is more than 1.8 times the mean sieve size of the size fraction is said to be
elongated. The mean size is defined as the arithmetic mean of the sieve size on which
the particle is just retained and the sieve size through which the particle just passes.
As closer size control is necessary, the sieves considered are not those of the standard
concrete aggregate series but: 75.0, 63.0, 50.0, 37.5, 28.0, 20.0, 14.0, 10.0, and 6.30

mm (or about 3, , 2, , 1, , , , and in.) sieves. The flakiness and elongation

tests are useful for general assessment of aggregates but they do not adequately
describe the particle shape.
The presence of elongated particles in excess of 10 to 15 per cent of the mass of
coarse aggregate is generally considered undesirable, but no recognized limits are laid
down. British Standard BS 882 : 1992 limits the flakiness index of the coarse
aggregate to 50 for natural gravel and to 40 for crushed or partially crushed coarse
aggregate. However, for wearing surfaces, lower values of the flakiness index are
required. Newer standards do not prescribe absolute limits on flakiness.
Surface texture of the aggregate affects its bond to the cement paste and also
influences the water demand of the mix, especially in the case of fine aggregate.
The classification of the surface texture is based on the degree to which the particle
surfaces are polished or dull, smooth or rough; the type of roughness has also to be
described. Surface texture depends on the hardness, grain size and pore characteristics
of the parent material (hard, dense and fine-grained rocks generally having smooth
fracture surfaces) as well as on the degree to which forces acting on the particle
surface have smoothed or roughened it. Visual estimate of roughness is quite reliable
but, in order to reduce misunderstanding, the classification of BS 812-1 : 1975, given
in Table 3.4, should be followed. This standard has been superseded by BS EN 12620 :
2002. There is no recognized method of measuring the surface roughness but
Wright’s approach3.2 is of interest: the interface between the particle and a resin in
which it is set is magnified, and the difference between the length of the profile and
the length of an unevenness line drawn as a series of chords is determined. This is
taken as a measure of roughness. Reproducible results are obtained, but the method is
laborious and is not widely used.

Table 3.4. Surface Texture of Aggregates (BS 812 : 1 : 1975, with Examples

125
 Another approach is to use a shape coefficient and a surface texture coefficient
evaluated from a Fourier series method which a priori assumes ranges of the
harmonic system and also of a modified total roughness coefficient.3.53 It is doubtful
whether this type of approach is useful in evaluating and comparing the wide range of
shapes and texture properties encountered in practice. Some other approaches are
reviewed by Ozol.3.65
It seems that the shape and surface texture of aggregate influence considerably the
strength of concrete. The flexural strength is more affected than the compressive
strength, and the effects of shape and texture are particularly significant in the case of
high strength concrete. Some data of Kaplan’s3.3 are reproduced in Table 3.5 but this
gives no more than an indication of the type of influence, as some other factors may
not have been taken into account. The full role of shape and texture of aggregate in
the development of concrete strength is not known, but possibly a rougher texture
results in a larger adhesive force between the particles and the cement matrix.
Likewise, the larger surface area of angular aggregate means that a larger adhesive
force can be developed.

Table 3.5. Average Relative Importance of the Aggregate Properties Affecting

the Strength of Concrete3.3

126
 The shape and texture of fine aggregate have a significant effect on the water
requirement of the mix made with the given aggregate. If these properties of fine
aggregate are expressed indirectly by its packing, i.e. by the percentage voids in a
loose condition (see p. 128), then the influence on the water requirement is quite
definite3.42 (see Fig. 3.3). The influence of the voids in coarse aggregate is less
definite.3.42

Fig. 3.3. Relation between void content of sand in a loose condition and the water
requirement of concrete made with the given sand3.42
Flakiness and the shape of coarse aggregate in general have an appreciable effect
on the workability of concrete. Figure 3.4, reproduced from Kaplan’s3.4 paper, shows
the pattern of the relation between the angularity of coarse aggregate and the
compacting factor of concrete made with it. An increase in angularity from minimum

127
to maximum would reduce the compacting factor by about 0.09 but, in practice,
clearly there can be no unique relation between the two factors because other
properties of aggregate also affect the workability. Kaplan’s experimental
results,3.4 however, do not confirm that the surface texture is a factor.

Fig. 3.4. The relation between the angularity number of aggregate and the
compacting factor of concrete made with the given aggregate3.4
Bond of aggregate
Bond between aggregate and cement paste is an important factor in the strength of
concrete, especially the flexural strength, but the nature of bond is not fully
understood. Bond is due, in part, to the interlocking of the aggregate and the hydrated
cement paste due to the roughness of the surface of the former. A rougher surface,
such as that of crushed particles, results in a better bond due to mechanical
interlocking; better bond is also usually obtained with softer, porous, and
mineralogically heterogeneous particles. Generally, texture characteristics which
permit no penetration of the surface of the particles are not conducive to good bond.
In addition, bond is affected by other physical and chemical properties of aggregate,
related to its mineralogical and chemical composition, and to the electrostatic
condition of the particle surface. For instance, some chemical bond may exist in the
case of limestone, dolomite,3.54 and possibly siliceous aggregates, and at the surface of
polished particles some capillary forces may develop. However, little is known about
these phenomena, and relying on experience is still necessary to predict the bond
between the aggregate and the surrounding hydrated cement paste. In any case, for
good development of bond, it is necessary that the aggregate surface be clean and free
from adhering day particles.
As concrete is a composite material, consisting of aggregate and of hydrated
cement paste matrix, the modulus of elasticity of each component influences
the modulus of the composite. The difference between the moduli has a considerable
influence on the bond of the aggregate.3.91
The determination of the quality of bond of aggregate is rather difficult and no
accepted tests exist. Generally, when bond is good, a crushed specimen of normal
strength concrete should contain some aggregate particles broken right through, in

128
addition to the more numerous ones pulled out from their sockets. An excess of
fractured particles, however, might suggest that the aggregate is too weak. Because it
depends on the strength of the hydrated cement paste as well as on the properties of
aggregate surface, bond strength increases with the age of concrete; it seems that the
ratio of bond strength to the strength of the hydrated cement paste increases with
age.3.43 Thus, providing it is adequate, the bond strength per se may not be a
controlling factor in the strength of ordinary concrete. However, in high strength
concrete, there is probably a tendency for the bond strength to be lower than the
tensile strength of the hydrated cement paste so that preferential failure in bond takes
place. Indeed, the interface between the aggregate and the surrounding cement paste is
of importance, if only because coarse aggregate represents a discontinuity and
introduces a wall effect.
Barnes et al.3.64 found plates of calcium hydroxide oriented parallel to the interface,
with C-S-H behind. Also, the interface zone is rich in the finer particles of cement and
has a higher water/cement ratio than the bulk of the cement paste. These observations
explain the particular role of silica fume in improving the strength of concrete (see
p. 672).
The problem of failure of concrete is discussed more fully in Chapter 6. What is
relevant at this stage is the path of cracks under high stress. In a homogeneous
material, a crack would be normal to the tension force causing the crack to open and
therefore be straight or nearly so. However, in a grossly heterogeneous material such
as concrete, the crack path may be affected by the presence of coarse aggregate. Thus
the crack may pass through the aggregate, or around the aggregate through the
interface zone, or through the matrix.
A recent study by Neville3.91 showed that, beyond a very early age, the crack path is
not influenced by the strength of the mortar matrix; the main factors are the strength
of the parent rock, and the shape and surface texture of the particles. However, no
simple and measurable parameters have been established. Nevertheless, the topic is of
practical interest with reference to the so-called aggregate interlock under shear.
Specifically, limestone aggregate fracture produces surfaces too smooth for shear
transfer.3.92
Strength of aggregate
Clearly, the compressive strength of concrete cannot significantly exceed that of
the major part of the aggregate contained therein, although it is not easy to state what
is the strength of the individual particles. Indeed, it is difficult to test the crushing
strength of individual aggregate particles, and the required information has to be
obtained usually from indirect tests: crushing value of bulk aggregate, force required
to compact bulk aggregate, and performance of aggregate in concrete.
The latter simply means either previous experience with the given aggregate or a
trial use of the aggregate in a concrete mix known to have a certain strength with
previously proven aggregates. If the aggregate under test leads to a lower compressive
strength of concrete, and in particular if numerous individual aggregate particles
appear fractured after the concrete specimen has been crushed, then the strength of the
aggregate is lower than the nominal compressive strength of the concrete mix in
which the aggregate was incorporated. Clearly, such aggregate can be used only in a
concrete of lower strength. This is, for instance, the case with laterite, a material
widely spread in Africa, South Asia and South America, which can rarely produce
concrete stronger than 10 MPa (1500 psi).

129
 Inadequate strength of aggregate represents a limiting case because the physical
properties of aggregate have some influence on the strength of concrete even when
the aggregate by itself is strong enough not to fracture prematurely. If we compare
concretes made with different aggregates, we can observe that the influence of
aggregate on the strength of concrete is qualitatively the same whatever the mix
proportions, and is the same regardless of whether the concrete is tested in
compression or in tension.3.5 It is possible that the influence of aggregate on the
strength of concrete is due not only to the mechanical strength of the aggregate but
also, to a considerable degree, to its absorption and bond characteristics.
In general, the strength and elasticity of aggregate depend on its composition,
texture, and structure. Thus, a low strength may be due to the weakness of constituent
grains or the grains may be strong but not well knit or cemented together.
The modulus of elasticity of aggregate is rarely determined; this is, however not
unimportant because the modulus of elasticity of concrete is generally higher the
higher the modulus of the constituent aggregate, but depends on other factors as well.
The modulus of elasticity of aggregate affects also the magnitude of creep and
shrinkage that can be realized by the concrete (see p. 453). A large incompatibility
between the moduli of elasticity of the aggregate and of the hydrated cement paste
adversely affects the development of microcracking at the aggregate–matrix interface.
A good average value of the crushing strength of aggregate is about 200 MPa (30
000 psi) but many excellent aggregates range in strength down to 80 MPa (12 000 psi).
One of the highest values recorded is 530 MPa (77 000 psi) for a certain quartzite.
Values for other rocks are given in Table 3.6.3.6 It should be noted that the required
strength of aggregate is considerably higher than the normal range of concrete
strengths because the actual stresses at the interface of individual particles within the
concrete may be far in excess of the nominal compressive stress applied.

Table 3.6. Compressive Strength of American Rocks Commonly Used as

Concrete Aggregates3.6

130
 On the other hand, aggregate of moderate or low strength and modulus of elasticity
can be valuable in preserving the integrity of concrete. Volume changes of concrete,
arising from hygral or thermal causes, lead to a lower stress in the hydrated cement
paste when the aggregate is compressible. Thus, compressibility of aggregate would
reduce distress in concrete while a strong and rigid aggregate might lead to cracking
of the surrounding cement paste.
It may be noted that no general relation exists between the strength and modulus of
elasticity of different aggregates.3.3 Some granites, for instance, have been found to
have a modulus of elasticity of 45 GPa (6.5 × 106 psi), and gabbro and diabase a
modulus of 85.5 GPa (12.4 × 106 psi), the strength of all these rocks ranging between
145 and 170 MPa (21 000 to 25 000 psi). Values of the modulus in excess of 160 GPa
(23 × 106 psi) have been encountered.
A test to measure the compressive strength of prepared rock cylinders used to be
prescribed. However, the results of such a test are affected by the presence of planes
of weakness in the rock which may not be significant once the rock has been
comminuted to the size used in concrete. In essence, the crushing strength test
measures the quality of the parent rock rather than the quality of the aggregates as
used in concrete. For this reason the test is rarely used.
Sometimes, the strength of a wet, as well as of a dry, specimen of rock is
determined. The ratio of wet to dry strengths measures the softening effect, and when
this is high, poor durability of the rock may be suspected.
A test on the crushing properties of bulk aggregate is the so-called crushing value
test of BS 812-110 : 1990, which measures resistance to pulverization.3.38 The crushing
value is a useful guide when dealing with aggregates of unknown performance,
particularly when lower strength may be suspected. There is no obvious physical

131
relation between this crushing value and the compressive strength, but the results of
the two tests are usually in agreement.3.75

The material to be tested for crushing value should pass a 14.0 mm ( in.) test

sieve and be retained on a 10.0 mm ( in.) sieve. When, however, this size is not
available, particles of other sizes may be used, but those larger than standard will in
general give a higher crushing value, and the smaller ones a lower value, than would
be obtained with the same rock of standard size. The sample to be tested should be
dried in an oven at 100 to 110 °C (212 to 230 °F) for four hours, and then placed in a
cylindrical mould and tamped in a prescribed manner. A plunger is put on top of the
aggregate and the whole assembly is placed in a compression testing machine and
subjected to a load of 400 kN (40 ton) (pressure of 22.1 MPa (3200 psi)) over the
gross area of the plunger, the load being increased gradually over a period of 10
minutes. After the load has been released, the aggregate is removed and sieved on a
2.36 mm (No. 8 ASTM*) test sieve in the case of a sample of the 14.0 to 10.0 mm

( to in.) standard size; for aggregates of other sizes, the sieve size is prescribed in
BS 812 : 110 : 1990. The ratio of mass of the material passing the smaller sieve to the
total mass of the sample is called the aggregate crushing value. Attempts to develop a
test for lightweight aggregates, similar to the crushing value test described above,
have been made but no test has been standardized.
*
For numbers of sieves see Table 3.14.
The crushing value test is rather insensitive to the variation in strength of weaker
aggregates, i.e. those with a crushing value greater than about 25. This is so because,
having been crushed before the full load of 400 kN (40 ton) has been applied, these
weaker materials become compacted so that the amount of crushing during later
stages of the test is reduced. Likewise, flaky particles increase the crushing
value.3.38 For these reasons, a ten per cent fines value test has been introduced and is
included in the BS 812-111 : 1990. In this test, the apparatus of the standard crushing
test is used to determine the load required to produce 10 per cent fines from the 14.0

to 10.0 mm ( to in.) particles compacted in a cylinder. This is achieved by

applying a progressively increasing load on the plunger so as to cause its penetration
in 10 minutes of about:
15 mm (0.60 in.) for rounded aggregate,
20 mm (0.80 in.) for crushed aggregate, and
24 mm (0.95 in.) for honeycombed aggregate (such as expanded shale or foamed
slag).
These penetrations should result in a percentage of fines passing a 2.36 mm (No. 8
ASTM) sieve of between 7.5 and 12.5 per cent. If y is the actual percentage of fines
due to a maximum load of x tons, then the load required to give 10 per cent fines is
given by 14x/(y + 4).
It should be noted that in this test, unlike the standard crushing value test, a higher
numerical result denotes a higher strength of the aggregate. British Standard BS 882 :
1992 (superseded by BS EN 12620 : 2002) prescribes a minimum ten per cent fines
value of 150 kN (15 ton) for aggregate to be used in heavy duty floors, 100 kN (10
ton) for concrete for wearing surfaces, and 50 kN (5 ton) when used in other concretes.

132
 The ten per cent fines value test shows a fairly good correlation with the standard
crushing value test for strong aggregates, while for weaker aggregates the ten per cent
fines value test is more sensitive and gives a truer picture of differences between more
or less weak aggregates. For this reason, the test is of use in assessing lightweight
aggregates but there is no simple relation between the test result and the upper limit of
strength of concrete made with the given aggregate.
Other mechanical properties of aggregate
Several mechanical properties of aggregate are of interest, especially when the
aggregate is to be used in pavement construction or is to be subjected to high wear.
The first of these is toughness, which can be defined as the resistance of a sample
of rock to failure by impact. Although this test would disclose adverse effects of
weathering of the rock, it is not used.
It is possible also to determine the impact value of bulk aggregate, and toughness
determined in this manner is related to the crushing value, and can, in fact, be used as
an alternative test. The size of the particles tested is the same as in the crushing value
test, and the permissible values of the crushed fraction smaller than a 2.36 mm (No. 8
ASTM) test sieve are also the same. The impact is applied by a standard hammer
falling 15 times under its own weight upon the aggregate in a cylindrical container.
This results in fragmentation in a manner similar to that produced by the pressure of
the plunger in the aggregate crushing value test. Details of the test are prescribed in
BS 812-112 : 1990 (2000), and BS 882 : 1992 prescribes the following maximum
values: 25 per cent when the aggregate is to be used in heavy duty floors; 30 per cent
when the aggregate is to be used in concrete for wearing surfaces; and 45 per cent
when it is to be used in other concretes. These figures serve as useful guides, but it is
clear that a direct correlation between the crushing value and the performance of
aggregate in concrete or the strength of the concrete is not possible.
One advantage of the impact value test is that it can be performed in the field with
some modifications, such as the measurement of quantities by volume rather than by
mass, but the test may not be adequate for compliance purposes.
In addition to strength and toughness, hardness or resistance to wear is an
important property of concrete used in pavements and in floor surfaces subjected to
heavy traffic. Several tests are available because it is possible to cause wear of
aggregate by abrasion, i.e. by rubbing of a foreign material against the stone under
test, or by attrition of stone particles against one another.
It may be worth noting that some limestone rocks are subject to wear, and their use
in concrete pavement should be conditional on abrasion testing. In other respects,
many limestone aggregates, even when porous, can produce satisfactory concrete.3.67
Abrasion of rock specimens is no longer determined and, in keeping with the
tendency to test aggregate in bulk, an abrasion value test on aggregate particles is
prescribed by BS 812-113 : 1990. Aggregate particles between 14.0 and 10.2 mm,
with flaky particles removed, are embedded in resin in a single layer. The sample is
subjected to abrasion in a standard machine, the grinding lap being turned 500
revolutions with Leighton Buzzard sand (see p. 54) being fed continuously at a
prescribed rate. The aggregate abrasion value is defined in terms of the percentage
loss in mass on abrasion, so that a high value denotes a low resistance to abrasion. BS
812-113 : 1990 has been withdrawn, and superseded by BS EN 12620 : 2002 where

133
the value of the coefficient has to be declared. PD 6682-1 : 2009 gives information
about abrasion tests.
European standard BS EN 12620 : 2002 prescribes the determination of a so-called
micro-Deval coefficient, which is a measure of wear of 10 to 14 mm aggregate
particles produced by friction between those particles and an abrasive charge in a
rotating drum. The coefficient represents the percentage loss in mass in the form of
particles reduced to a size smaller than 1.6 mm.
The attrition (Deval) test also uses aggregate in bulk but is no longer used because
it gives only small numerical differences between widely differing aggregates.
An American test combining attrition and abrasion is the Los Angeles test; it is
quite frequently used in other countries, too, because its results show good correlation
not only with the actual wear of aggregate when used in concrete but also with the
compressive and flexural strengths of concrete made with the given aggregate. In this
test, aggregate of specified grading is placed in a cylindrical drum, mounted
horizontally, with a shelf inside. A charge of steel balls is added, and the drum is
rotated a specified number of revolutions. The tumbling and dropping of the
aggregate and the balls results in abrasion and attrition of the aggregate, and this is
measured in the same way as in the attrition test.
The Los Angeles test can be performed on aggregates of different sizes, the same
wear being obtained by an appropriate mass of the sample and of the charge of steel
balls, and by a suitable number of revolutions. The various quantities are prescribed
by ASTM C 131-06. The Los Angeles test is, however, not very suitable for the
assessment of the behaviour of fine aggregate when subjected to attrition on
prolonged mixing; limestone fine aggregate is probably one of the more common
materials to undergo this degradation. For this reason, unknown fine aggregates
should, in addition to standard tests, be subjected to a wet attrition test to see how
much material smaller than 75 μm (No. 200 ASTM) sieve is produced. The degree to
which fine aggregate can be subject to degradation in the mixer can be determined by
the method of ASTM C 1137-05.
Table 3.7 gives average values of crushing strength, aggregate crushing value, and
abrasion, impact, and attrition values for the different rock groups of BS 812-1 : 1975
which has been superseded by BS EN 12620 : 2002. That standard deals also with
recycled aggregate, that is aggregate resulting from processing inorganic materials
previously used in construction. Recycled aggregate is not discussed in this book. It
should be noted that the values for hornfels and schists are based on a few specimens
only; these groups would appear to be better than they really are, presumably because
only good quality hornfels and schists were tested. As a rule, they are not suitable for
use in concrete. Likewise, chalk is not included in the limestone group data as it is not
generally suitable as a concrete aggregate.

Table 3.7. Average Test Values for British Rocks of Different Groups*

134
 As far as the crushing strength is concerned, basalt is extremely variable, fresh
basalts with little olivine reaching some 400 MPa (60 000 psi), while decomposed
basalt at the other end of the scale may have a strength of no more than 100 MPa (15
000 psi). Limestone and porphyry show much less variation in strength, and in Great
Britain porphyry has a good general performance – rather better than that of granites,
which tend to be variable.
An indication of the accuracy of the results of the different tests is given in Table
3.8, listing the number of samples to be tested in order to ensure a 90 per cent
probability that the mean value for the samples is within ±3 per cent and also within
±10 per cent of the true mean.3.40 The aggregate crushing value shows up as
particularly consistent. On the other hand, the prepared specimens show a greater
scatter of results than the bulk samples, which is of course to be expected. While the
various tests described in this and succeeding sections give an indication of the
quality of the aggregate, it is not possible to predict from the properties of aggregate
the potential strength development of concrete made with the given aggregate, and
indeed it is not yet possible to translate physical properties of aggregate into its
concrete-making properties.

Table 3.8. Reproducibility of Test Results on Aggregate3.40 (Crown copyright)

135
Specific gravity
Because aggregate generally contains pores, both permeable and impermeable (see
p. 129), the meaning of the term specific gravity has to be carefully defined, and there
are indeed several types of specific gravity.
The absolute specific gravity refers to the volume of the solid material excluding
all pores, and can, therefore, be defined as the ratio of the mass of the solid, referred
to vacuum, to the mass of an equal volume of gas-free distilled water, both taken at a
stated temperature. Thus, in order to eliminate the effect of totally enclosed
impermeable pores the material has to be pulverized, and the test is both laborious and
sensitive. Fortunately, it is not normally required in concrete technology work.
If the volume of the solid is deemed to include the impermeable pores, but not the
capillary ones, the resulting specific gravity is prefixed by the word apparent. The
apparent specific gravity is then the ratio of the mass of the aggregate dried in an oven
at 100 to 110 °C (212 to 230 °F) for 24 hours to the mass of water occupying a
volume equal to that of the solid including the impermeable pores. The latter mass is
determined using a vessel which can be accurately filled with water to a specified
volume. Thus, if the mass of the oven-dried sample is D, the mass of the vessel full of
water is B, and the mass of the vessel with the sample and topped up with water is A,
then the mass of the water occupying the same volume as the solid is B – (A – D). The
apparent specific gravity is then

The vessel referred to earlier, and known as a pycnometer, is usually a one-litre jar
with a watertight metal conical screwtop having a small hole at the apex. The
pycnometer can thus be filled with water so as to contain precisely the same volume
every time.
Calculations with reference to concrete are generally based on the saturated and
surface-dry condition of the aggregate (see p. 132) because the water contained
in all the pores in the aggregate does not take part in the chemical reactions of cement

136
and can, therefore, be considered as part of the aggregate. Thus, if a sample of the
saturated and surface-dry aggregate has a mass C, the gross apparent specific gravity
is

This is the specific gravity most frequently and easily determined, and it is used for
calculations of yield of concrete or of the quantity of aggregate required for a given
volume of concrete.
The apparent specific gravity of aggregate depends on the specific gravity of the
minerals of which the aggregate is composed and also on the amount of voids. The
majority of natural aggregates have a specific gravity of between 2.6 and 2.7, and the
range of values is given in Table 3.9.3.7 The values for artificial aggregates extend
from considerably below to very much above this range (see Chapter 13).

Table 3.9. Apparent Specific Gravity of Different Rock Groups3.7 (Crown

copyright)

As mentioned earlier, specific gravity of aggregate is used in the calculation of

quantities but the actual value of the specific gravity of aggregate is not a measure of
its quality. Thus, the value of specific gravity should not be specified unless we are
dealing with a material of a given petrological character when a variation in specific
gravity would reflect the porosity of the particles. An exception to this is the case of
mass construction, such as a gravity dam, where a minimum density of concrete is
essential for the stability of the structure.
Bulk density
It is well known that in the SI system the density of a material is numerically equal to
its specific gravity although, of course, the latter is a ratio while density is expressed
in kilograms per litre. However, in concrete practice, expressing the density of
kilograms per cubic metre is more common. In the American or Imperial system,
specific gravity has to be multiplied by the unit mass of water (approximately 62.4
lb/ft3) in order to be converted into absolute density (specific mass) expressed in
pounds per cubic foot.
This absolute density, it must be remembered, refers to the volume of the
individual particles only, and of course it is not physically possible to pack these
particles so that there are no voids between them. When aggregate is to be actually

137
batched by volume it is necessary to know the mass of aggregate that would fill a
container of unit volume. This is known as the bulk density of aggregate, and this
density is used to convert quantities by mass to quantities by volume.
The bulk density clearly depends on how densely the aggregate is packed, and it
follows that, for a material of a given specific gravity, the bulk density depends on the
size distribution and shape of the particles: particles all of one size can be packed to a
limited extent, but smaller particles can be added in the voids between the larger ones,
thus increasing the bulk density of the packed material. The shape of the particles
greatly affects the closeness of packing that can be achieved.
For a coarse aggregate of given specific gravity, a higher bulk density means that
there are fewer voids to be filled by fine aggregate and cement, and the bulk density
test was at one time used as a basis of proportioning of mixes.
The actual bulk density of aggregate depends not only on the various
characteristics of the material which determine the potential degree of packing, but
also on the actual compaction achieved in a given case. For instance, using spherical
particles all of the same size, the densest packing is achieved when their centres lie at
the apexes of imaginary tetrahedra. The bulk density is then 0.74 of the absolute
density (specific mass) of the material. For the loosest packing, the centres of spheres
are at the corners of imaginary cubes and the bulk density is only 0.52 of the specific
mass of the solid.
Thus, for test purposes, the degree of compaction has to be specified. British
Standard BS 812 : 2 : 1995 recognizes two degrees: loose (or uncompacted) and
compacted. The test is performed in a metal cylinder of prescribed diameter and depth,
depending on the maximum size of the aggregate and also on whether compacted or
uncompacted bulk density is being determined.
For the determination of loose bulk density, the dried aggregate is gently placed in
the container to overflowing and then levelled by rolling a rod across the top. In order
to find the compacted or rodded bulk density, the container is filled in three stages,
each third of the volume being tamped a prescribed number of times with a 16 mm
( in.) diameter round-nosed rod. Again, the overflow is removed. The net mass of the
aggregate in the container divided by its volume then represents the bulk density for
either degree of compaction. The ratio of the loose bulk density to the compacted bulk
density lies usually between 0.87 and 0.96.3.55
Knowing the apparent specific gravity for the saturated and surface-dry
condition, s, the voids ratio can be calculated from the expression:

If the aggregate contains surface water, it will pack less densely owing to the
bulking effect. This is discussed on p. 134. Moreover, the bulk density as determined
in the laboratory may not be directly suitable for conversion of mass to volume of
aggregate for purposes of volume batching as the degree of compaction in the
laboratory and on the site may not be the same.
The bulk density of aggregate is of interest in connection with the use of
lightweight and heavy aggregates (see p. 763).
Porosity and absorption of aggregate

138
The presence of internal pores in the aggregate particles was mentioned in connection
with the specific gravity of aggregate, and indeed the characteristics of these pores are
very important in the study of its properties. The porosity of aggregate, its
permeability, and absorption influence such properties of aggregate as the bond
between it and the hydrated cement paste, the resistance of concrete to freezing and
thawing, as well as its chemical stability and resistance to abrasion. As stated earlier,
the apparent specific gravity of aggregate also depends on its porosity and, as a
consequence, the yield of concrete for a given mass of aggregate is affected (see
p. 764).
The pores in aggregate vary in size over a wide range, the largest being large
enough to be seen under a microscope or even with the naked eye, but even the
smallest aggregate pores are larger than the gel pores in the cement paste. Pores
smaller than 4 μm are of special interest as they are generally believed to affect the
durability of aggregates subjected to alternating freezing and thawing (see 545).
Some of the aggregate pores are wholly within the solid; others open onto the
surface of the particle. The cement paste, because of its viscosity, cannot penetrate to
a great depth any but the largest of the aggregate pores, so that is the gross volume of
the particle that is considered solid for the purpose of calculating the aggregate
content in concrete. However, water can enter the pores, the amount and rate of
penetration depending on their size, continuity and total volume. The values of
porosity of some common rocks are given in Table 3.10 and, since aggregate
represents some three-quarters of the volume of concrete, it is clear that the porosity
of aggregate materially contributes to the overall porosity of concrete.

Table 3.10. Porosity of Some Common Rocks

When all the pores in the aggregate are full, it is said to be saturated and surface-
dry. If aggregate in this condition is allowed to stand free in dry air, e.g. in the
laboratory, some of the water contained in the pores will evaporate and the aggregate
will be less than saturated, i.e. air-dry. Prolonged drying in an oven would reduce the
moisture content of the aggregate still further until, when no moisture whatever is left,
the aggregate is said to be bone-dry. These various stages are shown diagrammatically
in Fig. 3.5, and some typical values of absorption are given in Table 3.11. At the
extreme right of Fig. 3.5, the aggregate contains surface moisture and is darker in
colour.

Table 3.11. Typical Values of Absorption of Different British Aggregates3.8

139
 Fig. 3.5. Diagrammatic representation of moisture in aggregate
The water absorption of aggregate is determined by measuring the increase in mass
of an oven-dried sample when immersed in water for 24 hours (the surface water
being removed). The ratio of the increase in mass to the mass of the dry sample,
expressed as a percentage, is termed absorption. Standard procedures are prescribed in
BS 812-2 : 1995.
Some typical values of absorption of different aggregates are given in Table 3.11,
based on Newman’s data.3.8 The moisture content in the air-dry condition is also
tabulated. It may be noted that gravel has generally a higher absorption than crushed

140
rock of the same petrological character because weathering results in the outer layer
of the gravel particles being more porous and absorbent.
Although there is no clear-cut relation between the strength of concrete and the
water absorption of aggregate used, the pores at the surface of the particle affect the
bond between the aggregate and the cement paste, and may thus exert some influence
on the strength of concrete.
Normally, it is assumed that, at the time of setting of concrete, the aggregate is in a
saturated and surface-dry condition. If the aggregate is batched in a dry condition, it is
assumed that sufficient water will be absorbed from the mix to bring the aggregate to
a saturated condition, and this absorbed water is not included in the free or effective
mixing water. Such a situation can be encountered in a hot, dry climate. It is possible,
however, that, when dry aggregate is used, the particles become quickly coated with
cement paste which prevents further ingress of water necessary for saturation. This is
particularly so with coarse aggregate, where water has further to travel from the
surface of the particle. As a result, the effective water/cement ratio is higher than
would be the case had full absorption of water by the aggregate been possible. This
effect is significant mainly in rich mixes where rapid coating of aggregate can take
place; in lean, wet mixes the saturation of aggregate proceeds undisturbed. In practical
cases, the actual behaviour of the mix is affected also by the order of feeding the
ingredients into the mixer.
The absorption of water by aggregate results also in some loss of workability with
time, but beyond about 15 minutes the loss becomes small.
Because absorption of water by dry aggregate slows down or is stopped owing to
the coating of particles with cement paste, it is often useful to determine the quantity
of water absorbed in 10 to 30 minutes instead of the total water absorption, which
may never be achieved in practice.
Moisture content of aggregate
It was mentioned in connection with the specific gravity that, in fresh concrete, the
volume occupied by the aggregate is the volume of the particles including all the
pores. If no water movement into the aggregate is to take place, the pores must be full
of water, i.e. the aggregate must be in a saturated condition. On the other hand, any
water on the surface of the aggregate will contribute to the water in the mix and will
occupy a volume in excess of that of the aggregate particles. The basic state of the
aggregate is thus saturated and surface-dry.
Aggregate exposed to rain collects a considerable amount of moisture on the
surface of the particles and, except at the surface of the stockpile, keeps this moisture
over long periods. This is particularly true of fine aggregate, and the surface- or free
moisture (in excess of that held by aggregate in a saturated and surface-dry condition)
must be allowed for in the calculation of batch quantities. Coarse aggregate rarely
contains more than 1 per cent of surface moisture but fine aggregate can contain in
excess of 10 per cent. The surface moisture is expressed as a percentage of the mass
of the saturated and surface-dry aggregate, and is termed moisture content.
Since absorption represents the water contained in aggregate in a saturated and
surface-dry condition, and the moisture content is the water in excess of that state, the
total water content of a moist aggregate is equal to the sum of absorption and moisture
content.

141
 As the moisture content of aggregate changes with weather, and changes also from
one part of a stockpile to another, the value of the moisture content has to be
determined frequently and a number of methods have been developed. The oldest one
consists simply of finding the loss in mass of an aggregate sample when dried on a
tray over a source of heat. Care is necessary to avoid over-drying: the sand should be
brought to a just free-flowing condition, and must not be heated further. This stage
can be determined by feel or by forming the sand into a pile by means of a conical
mould; when the mould has been removed the material should slump freely. If the
sand has acquired a brownish tinge, this is a sure sign that excessive drying has taken
place. This method of determining the moisture content of aggregate, colloquially
referred to as the ‘frying-pan method’, is simple, can be used in the field, and is quite
reliable. Microwave ovens can also be used but care is necessary to avoid overheating.
In the laboratory, the moisture content of aggregate can be determined by means of
a pycnometer. The apparent specific gravity of the aggregate on a saturated and
surface-dry basis, s, must be known. Then, if B is the mass of the pycnometer full of
water, C the mass of the moist sample, and A the mass of the pycnometer with the
sample and topped up with water, the moisture content of the aggregate is:

The test is slow and requires great care in execution (e.g. all air must be expelled
from the sample) but can yield accurate results. This method is described in BS 812-
109 : 1990.
In the siphon can test3.9 the volume of water displaced by a known mass of moist
aggregate is measured, the siphon making this determination more accurate.
Preliminary calibration for each aggregate is required because the results depend on
its specific gravity but, once this has been done, the test is rapid and accurate.
The moisture content of aggregate can also be found using a steelyard moisture
meter: the moist aggregate is added to a vessel containing a fixed amount of water and
suspended at one end of a steelyard until it balances. We measure thus the quantity of
water that has to be replaced by the moist aggregate for a constant weight
and total volume. For this condition, it can be shown that the amount of displaced
water is proportional to the moisture content of the aggregate. A calibration curve for
any aggregate used has to be obtained. The moisture content can be determined with
an accuracy of per cent.
In the buoyancy meter test,3.10 the moisture content of the aggregate of known
specific gravity is determined from the apparent loss in weight on immersion in water.
The balance can read the moisture content direct if the size of the sample is adjusted,
according to the specific gravity of the aggregate, to such a value that a saturated and
surface-dry sample has a standard weight when immersed. The test is rapid and gives
the moisture content to the nearest per cent. A simple version of the test is
prescribed by ASTM C 70-06 but is not widely used.
Numerous other methods have been developed. For instance, moisture can be
removed by burning the aggregate with methyl alcohol, the resulting loss in mass of
the sample being measured. There are also proprietary meters based on the
measurement of pressure of gas formed in a closed vessel by the reaction of calcium
carbide with the moisture in the sample. ASTM C 566-97 (2004) prescribes a method

142
for the determination of the total moisture content of the aggregate. Although this
method is not highly accurate, the error involved is smaller than the sampling error.
It can be seen that a great variety of tests is available but, however accurate the test,
its result is useful only if a representative sample has been used. Furthermore, if the
moisture content of aggregate varies between adjacent parts of a stockpile, the
adjustment of mix proportions becomes laborious. Because the variation in moisture
content occurs mainly in the vertical direction from a waterlogged bottom of a pile to
its drying or dry surface, care in laying out of stockpiles is necessary: storing in
horizontal layers, having at least two stockpiles and allowing each pile to drain before
use, and not using the bottom 300 mm (12 in.) or so, all help to keep the variation in
moisture content to a minimum. Coarse aggregate holds very much less water than
fine aggregate, has a less variable moisture content, and generally causes fewer
difficulties.
Electrical devices which give instantaneous or continuous reading of the moisture
content of aggregate in a storage bin, on the basis of the variation of resistance or
capacitance with a change in the moisture content of the aggregate, have been
developed. In some batching plants, meters of this type are used in automatic devices
which regulate the quantity of water to be added to the mixer but an acuracy greater
than 1 per cent of moisture cannot in practice be achieved; frequent calibration is
necessary. Measurement of the dielectric constant has the advantage of not being
affected by the presence of salts. Microwave absorption meters have been developed:
these are accurate and stable but expensive. Instruments emitting fast neutrons which
are thermalized by hydrogen atoms in the water are also used. All these instruments
need to be carefully positioned.
There is no doubt that continuous measurement of moisture and automatic
adjustment of the amount of water added into the mixer greatly reduce the variability
of the concrete produced when the moisture content of the aggregate is variable.
However, widespread adoption of the determination of the moisture content in the
aggregate in any given batch is still in the future.
Bulking of fine aggregate
The presence of moisture in aggregate necessitates correction of the actual mix
proportions: the mass of water added to the mix has to be decreased by the mass of
the free moisture in the aggregate, and the mass of the wet aggregate must be
increased by a like amount. In the case of sand, there is a second effect of the
presence of moisture: bulking. This is the increase in the volume of a given mass of
sand caused by the films of water pushing the sand particles apart. While bulking per
se does not affect the proportioning of materials by mass, in the case of volume
batching, bulking results in a smaller mass of sand occupying the fixed volume of the
measuring box. For this reason, the mix becomes deficient in fine aggregate and
appears ‘stony’, and the concrete may be prone to segregation and honeycombing.
Also, the yield of concrete is reduced. The remedy, of course, lies in increasing the
apparent volume of fine aggregate (sand) to allow for bulking.
The extent of bulking depends on the percentage of moisture present in the sand
and on its fineness. The increase in volume relative to that occupied by a saturated
and surface-dry sand increases with an increase in the moisture content of the sand up
to a value of some 5 to 8 per cent, when the bulking of 20 to 30 per cent occurs. Upon
further addition of water, the films merge and the water moves into the voids between
the particles so that the total volume of sand decreases until, when fully saturated

143
(flooded), its volume is approximately the same as the volume of dry sand for the
same method of filling the container. This is apparent from Fig. 3.6, which also shows
that finer sand bulks considerably more and reaches maximum bulking at a higher
water content than does coarse sand. Crushed fine aggregate bulks even more than
natural sand. Extremely fine sand (which contains a larger number of particles) has
been known to bulk as much as 40 per cent at a moisture content of 10 per cent, but
such a sand is, in any case, not used for the production of good quality concrete.

Fig. 3.6. Decrease in true volume of sand due to bulking (for a constant apparent
volume of moist sand)
Coarse aggregate shows only a negligible increase in volume due to the presence
of free water, as the thickness of moisture films is very small compared with the
particle size.
Because the volume of saturated sand is the same as that of dry sand, the most
convenient way of determining bulking is by measuring the decrease in volume of the
given sand when inundated. A container of known volume is filled with loosely
packed moist sand. The sand is then tipped out, the container is partially filled with
water and the sand is gradually fed back, with stirring and rodding to expel all air
bubbles. The volume of sand in the saturated state, Vs, is now measured. If Vm is the
initial apparent volume of the sand (i.e. the volume of the container), then bulking is
given by (Vm – Vs)/Vs.
With volume batching, bulking has to be allowed for by increasing the total
volume of (moist) sand used. Thus, volume Vs is multiplied by the factor:

sometimes known as the bulking factor; a plot of bulking factor against moisture of
three typical sands is shown in Fig. 3.7.

144
 Fig. 3.7. Bulking factor for sands with different moisture contents
The bulking factor can also be found from the bulk densities of dry and moist
sand, Dd and Dm, respectively, and the moisture content per unit volume of sand, m/Vm.
The bulking factor is then

Since Dd represents a ratio of the mass of dry sand, w, to its bulk volume Vs (the
volumes of dry and inundated sand being the same),

i.e. the two factors are identical.

Deleterious substances in aggregate
There are three broad categories of deleterious substances that may be found in
aggregates: impurities which intefere with the processes of hydration of
cement; coatings preventing the development of good bond between aggregate and
the hydrated cement paste; and certain individual particles which
are weak or unsound in themselves. All or part of an aggregate can also be harmful
through the development of chemical reactions between the aggregate and the cement
paste: these chemical reactions are discussed on p. 144.
Organic impurities
Natural aggregates may be sufficiently strong and resistant to wear and yet they may
not be satisfactory for concrete-making if they contain organic impurities which
interfere with the chemical reactions of hydration. The organic matter found in
aggregate consists usually of products of decay of vegetable matter (mainly tannic
acid and its derivatives) and appears in the form of humus or organic loam. Such
materials are more likely to be present in sand than in coarse aggregate, which is
easily washed.
Not all organic matter is harmful and it is best to check its effects by making actual
compression test specimens. Generally, however, it saves time to ascertain first
whether the amount of organic matter is sufficient to warrant further tests. This is

145
done by the so-called colorimetric test of ASTM C 40-04. The acids in the sample are
neutralized by a 3 per cent solution of NaOH, prescribed quantities of aggregate and
of solution being placed in a bottle. The mixture is vigorously shaken to allow the
intimate contact necessary for chemical reaction, and then left to stand for 24 hours,
when the organic content can be judged by the colour of the solution: the greater the
organic content the darker the colour. If the colour of the liquid above the test sample
is not darker than the standard yellow colour defined by the ASTM Standard, the
sample can be assumed to contain only a harmless amount of organic impurities.
If the observed colour is darker than the standard yellow, i.e. if the solution
appears brownish or brown, the aggregate has a rather high organic content, but this
does not necessarily mean that the aggregate is not fit for use in concrete. The organic
matter present may not be harmful to concrete or the colour may be due to some iron-
bearing minerals. For this reason, further tests are necessary: ASTM C 87-05
recommends strength tests on mortar with the suspect sand as compared with mortar
made with the same, but washed, sand. The colorimetric test is no longer specified in
British Standards.
In some countries, the quantity of organic matter in aggregate is determined from
the loss of mass of a sample on treating with hydrogen peroxide.
It is interesting to note that, in some cases, the effects of organic impurities may be
only temporary. In one investigation,3.11 concrete made with a sand containing organic
matter had a 24-hour strength equal to 53 per cent of the strength of similar concrete
made with clean sand. At 3 days, this ratio rose to 82 per cent, then to 92 per cent at 7
days, and at 28 days equal strengths were recorded.
Clay and other fine material
Clay may be present in aggregate in the form of surface coatings which interfere with
the bond between aggregate and the cement paste. Because good bond is essential to
ensure a satisfactory strength and durability of concrete, the problem of clay coatings
is an important one.
There are two more types of fine material which can be present in aggregate: silt
and crusher dust. Silt is material between 2 and 60 μm, reduced to this size by natural
processes of weathering; silt may thus be found in aggregate won from natural
deposits. On the other hand, crusher dust is a fine material formed during the process
of comminution of rock into crushed stone or, less frequently, of gravel into crushed
fine aggregate. In a properly laid out processing plant, this dust should be removed by
washing. Other soft or loosely adherent coatings can also be removed during the
processing of the aggregate. Well-bonded coatings cannot be so removed but, if they
are chemically stable and have no deleterious effect, there is no objection to the use of
aggregate with such a coating, although shrinkage may be increased. However,
aggregates with chemically reactive coatings, even if physically stable, can lead to
serious trouble.
Silt and fine dust may form coatings similar to those of clay, or may be present in
the form of loose particles not bonded to the coarse aggregate. Even when they are in
the latter form, silt and fine dust should not be present in excessive quantities because,
owing to their fineness and therefore large surface area, silt and fine dust increase the
amount of water necessary to wet all the particles in the mix.
In view of the above, it is necessary to control the clay, silt and fine dust contents
of aggregate. As no test is available to determine separately the clay content, this is

146
not prescribed in British Standards. However, BS 882 : 1992 (withdrawn 2004 and
replaced by BS EN 12620 : 2002 and PD 6682-1 : 2009) imposes a limit on the
maximum amount of material passing the 75 μm (No. 200) sieve:
in coarse aggregate: 2 per cent, increased to 4 per cent when it consists wholly of
crushed rock;
in fine aggregate: 4 per cent, increased to 16 per cent when it consists wholly of
crushed rock; and
in all-in aggregate: 11 per cent.
For heavy duty floors, the limit is 9 per cent. ASTM 33-08 also specifies grading
requirements. European standard BS EN 12620 : 2002 requires the content of fines to
be declared.
In the BS standard, the content of clay lumps and friable particles is specified
separately as 3 per cent in fine, and 2 to 10 per cent in coarse aggregate, depending on
the use of the concrete.
It should be noted that different test methods are prescribed in different standards
so that the results are not directly comparable.
The clay, silt and fine dust content of fine aggregate can be determined by the
sedimentation method described in BS 812-103.2 : 1989 (2000). The sample is placed
in a sodium hexametaphosphate solution in a stoppered jar and rotated with the axis of
the jar horizontal for 15 minutes at approximately 80 revolutions per minute. The fine
solids become dispersed and the amount of suspended material is then measured by
means of a pipette. A simple calculation gives the percentage of clay, fine silt and fine
dust in the fine aggregate, the separation size being 20 μm.
A similar method, with suitable modifications, can be used for coarse aggregate
which contains very fine material, but it is simpler to wet-sieve the aggregate on a
75 μm (No. 200) test sieve, as prescribed in BS 812 : 103.1 : 1985 (2000) and ASTM
C 117-04. This type of sieving is resorted to because fine dust or clay adhering to
larger particles would not be separated in ordinary dry sieving. In wet sieving, on the
other hand, the aggregate is placed in water and agitated sufficiently vigorously for
the finer material to be brought into suspension. By decantation and sieving, all
material smaller than a 75 μm (No. 200) test sieve can be removed. To protect this
sieve from damage by large particles during decantation, a 1.18 mm (No. 16 ASTM)
sieve is placed above the 75 μm (No. 200) sieve.
For natural sands and crushed gravel sands, there is also a field test available
which can be performed quite easily and rapidly, with very little laboratory equipment.
In this non-standard test, 50 ml of an approximately 1 per cent solution of common
salt in water is placed in a 250 ml measuring cylinder. Sand, as received, is added
until its level reaches the 100 ml mark, and more solution is then added until the total
volume of the mixture in the cylinder is 150 ml. The cylinder is now covered with the
palm of the hand, shaken vigorously, repeatedly turned upside down, and then
allowed to stand for 3 hours. The silt which became dispersed on shaking will now
settle in a layer above the sand, and the height of this layer can be expressed as a
percentage of the height of the sand below.
It should be remembered that this is a volumetric ratio, which cannot easily be
converted to a ratio by mass since the conversion factor depends on the fineness of the
material. It has been suggested that for natural sand the mass ratio is obtained by
multiplying the volumetric ratio by a factor of , the corresponding figure for crushed

147
gravel sand being , but with some aggregates an even wider variation is obtained.
These conversions are not reliable, so that, when the volumetric content exceeds 8 per
cent, tests by the more accurate methods, described earlier, should be made.
Salt contamination
Sand won from the seashore or dredged from the sea or a river estuary, as well as
desert sand, contains salt, and has to be processed. In the United Kingdom, about 20
per cent of natural gravel and sand is sea dredged, submersible pumps making it
possible to win the material from depths up to 50 m (160 ft). The simplest procedure
is to wash the sand in fresh water, but special care is required with deposits just above
the high-water mark in which large quantities of salt, sometimes over 6 per cent of
mass of sand, can be found. Generally, sand from the sea bed, washed even in sea
water, does not contain harmful quantities of salts.
Because of the danger of chloride-induced corrosion of steel reinforcement, BS
8110-1 : 1997 (Structural use of concrete) specifies the maximum total chloride ion
content in the mix. The chlorides may arise from all ingredients of the mix. As far as
aggregate is concerned, BS 882 : 1992 contains guidance on the maximum chloride
ion content in the aggregate which is likely to be acceptable, although the total
content in the concrete mix should be verified. The BS 882 : 1992 (withdrawn) limits
on the chloride ion content by mass, expressed as a percentage of the mass of the total
aggregate, are as follows:

The method of BS 812-117 : 1988 (2000) determines the content of water-soluble

chlorides; this may be inadequate when the aggregate is porous and chlorides may
exist within the aggregate particles.3.38
Apart from the danger of corrosion of steel reinforcement, if salt is not removed, it
will absorb moisture from the air and cause efflorescence – unsightly white deposits
on the surface of the concrete (see p. 515).
Sea-dredged coarse aggregate may have a large shell content. This usually has no
adverse effect on strength but workability of concrete made with aggregate having a
large shell content is slightly reduced.3.44 The shell content of particles larger than 5
mm can be determined by hand picking, using the method of BS 812-106 : 1985
replaced by BS EN 933-7 : 1998. British Standard BS 882 : 1992 (withdrawn) limits
the shell content of coarse aggregate to 20 per cent when the maximum size is 10 mm
and to 8 per cent when it is larger. Nevertheless, aggregate with a much higher shell
content has been used successfully on some Pacific islands. There is no limit on the
shell content of fine aggregate. The current standard is BS EN 12620 : 2002.
Unsound particles
Tests on aggregate sometimes reveal that the majority of the component particles are
satisfactory but that a few are unsound: the quantity of such particles must clearly be
limited.
There are two broad types of unsound particles: those that fail to maintain their
integrity, and those that lead to disruptive expansion on freezing or even on exposure

148
to water. The disruptive properties are characteristic of certain rock groups, and will
therefore be discussed in relation to the durability of aggregate in general (mainly in
the next section). In this section, non-durable impurities only will be considered.
Shale and other particles of low density are regarded as unsound, and so are soft
inclusions, such as clay lumps, wood, and coal as they lead to pitting and scaling. If
present in large quantities (over 2 to 5 per cent of the mass of the aggregate) these
particles may adversely affect the strength of concrete and should certainly not be
permitted in concrete which is exposed to abrasion.
Coal, in addition to being a soft inclusion, is undesirable for other reasons: it can
swell, causing disruption of concrete and, if present in large quantities in a finely
divided form, it can disturb the process of hydration of the cement paste. However,
discrete particles of hard coal amounting to no more than per cent of the mass of the
aggregate have no adverse effect on the strength of concrete.
The presence of coal and other materials of low density can be determined by
flotation in a liquid of suitable specific gravity, for instance, by the method of ASTM
C 123-04. If the danger of pitting and scaling is not thought important, and strength of
concrete is the main consideration, a trial mix should be made.
Mica should be avoided because, in the presence of active chemical agents
produced during the hydration of cement, alteration of mica to other forms can occur.
Also, free mica in fine aggregate, even in quantities of a few per cent of the mass of
the aggregate, affects adversely the water requirement and the strength of
concrete.3.45 Fookes and Revie3.69 found that a 5 per cent content by mass of mica in
sand reduced the 28-day strength of the resulting concrete by about 15 per cent, even
when the water/cement ratio was kept constant. The reason for this is probably poor
adhesion of the cement paste to the surface of mica particles. It appears that mica in
the form of muscovite is much more harmful than biotite.3.58 These facts should be
borne in mind when materials such as china clay sand are considered for use in
concrete.
There is no standard method of determining the amount of mica present in sand or
even a test on the effect of mica on the properties of concrete. If sand contains mica, it
is likely to be concentrated among the finest particles. Gaynor and
Meininger3.63 recommend a microscopical count of mica particles in the fraction of
sand between 300 and 150 μm (No. 50 to No. 100) sieve sizes, and if less than about
15 per cent of mica as a number of particles is present in that fraction, the properties
of concrete are unlikely to be significantly affected. It should be emphasized that the
mica content of the larger particles should be many times smaller.
Gypsum and other sulfates must not be present; their content in aggregate can be
determined by BS 812 : 118 : 1988. Other requirements for sulfate content are given
in PD 6682-1 : 2009. Their existence in many Middle East aggregates leads to
difficulties, but up to 5 per cent of SO3 by mass of cement (including the SO3 in the
cement) is often tolerated there.3.59 Water-soluble forms, such as magnesium and
sodium sulfates, are particularly harmful. Special problems arising from the presence
of various salts in aggregates found in arid regions, such as the Middle East, are
treated by Fookes and Collis.3.56,3.57
Iron pyrites and marcasite represent the most common expansive inclusions in
aggregate. These sulfides react with water and oxygen in the air to form a ferrous
sulfate which subsequently decomposes to form a hydroxide, while the sulfate ions
react with calcium aluminates in the cement. Sulfuric acid can also form, and this can

149
attack the hydrated cement paste.3.76 Surface staining of the concrete and disruption of
the cement paste (pop-outs) may result, particularly under warm and humid conditions.
The formation of pop-outs can be delayed for many years until water and oxygen are
present.3.76 The appearance problem of pop-outs can be ameliorated by using a smaller
maximum size of aggregate.
Not all forms of pyrites are reactive and, because the decomposition of pyrites
takes place only in lime water, it is possible to test a suspect aggregate for reactivity
by placing the material in a saturated solution of lime.3.12 If the aggregate is reactive, a
blue-green gelatinous precipitate of ferrous sulfate appears within a few minues and,
on exposure to air, this changes to brown ferric hydroxide. The absence of this
reaction means that no staining need be feared. Lack of reactivity was found3.12 to be
associated with the presence of a number of metal cations, while their absence makes
the pyrites active. Generally, particles of pyrites likely to cause trouble are those

between 5 and 10 mm (or and in.) in size.

The permissible quantities of unsound particles laid down by ASTM C 33-08 are
summarized in Table 3.12.

Table 3.12. Permissible Quantities of Unsound Particles Prescribed by ASTM C

33-08

The majority of impurities discussed in the present section are found in natural
aggregate deposits and are much less frequently encountered in crushed aggregate.
However, some processed aggregates, such as mine tailings, can contain harmful
substances. For instance, small quantities of lead soluble in limewater (e.g. 0.1 per
cent of PbO by mass of aggregate) greatly delay the set and reduce the early strength
of concrete; the long-term strength is unaffected.3.46
Soundness of aggregate
This is the term used to describe the ability of aggregate to resist excessive changes in
volume as a result of changes in physical conditions. Lack of soundness is thus
distinct from expansion caused by the chemical reactions between the aggregate and
the alkalis in cement.
The physical causes of large or permanent volume changes of aggregate are
freezing and thawing, thermal changes at temperatures above freezing, and alternating
wetting and drying.

150
 Aggregate is said to be unsound when volume changes, induced by the above
causes, result in deterioration of the concrete. This may range from local scaling and
so-called pop-outs to extensive surface cracking and to disintegration over a
considerable depth, and can thus vary from no more than impaired appearance to a
structurally dangerous situation.
Unsoundness is exhibited by porous flints and cherts, especially the lightweight
ones with a fine-textured pore structure, by some shales, by limestones with laminae
of expansive clay, and by other particles containing clay minerals, particularly of the
montmorillonite or illite group. For instance, an altered dolerite has been found to
change in dimensions by as much as 600 × 10–6 with wetting and drying, and concrete
containing this aggregate might fail under conditions of alternating wetting and drying,
and will certainly do so on freezing and thawing. Likewise, the disruption of porous
chert arises from freezing.3.77
A British test for soundness of aggregate is prescribed in BS 812-121 : 1989
(2000). This determines the proportion of aggregate broken up in consequence of five
cycles of immersion in a saturated solution of magnesium sulfate alternating with
oven drying. The original sample contains particles between 10.0 and 14.0 mm in size,
and the mass of particles which remain larger than 10.0 mm, expressed as a
percentage of the original mass, is called the soundness value.
The American test for soundness of aggregate is prescribed by ASTM C 88-05. A
sample of graded aggregate is subjected alternately to immersion in a saturated
solution of sodium or magnesium sulfate (the latter being the more severe of the two)
and drying in an oven. The formation of salt crystals in the pores of the aggregate
tends to disrupt the particles, probably in a manner similar to the action of ice. The
reduction in size of the particles, as shown by a sieve analysis, after a number of
cycles of exposure denotes the degree of unsoundness. The test is no more than
qualitative in predicting the behaviour of the aggregate under actual site conditions,
and cannot be used as a basis of acceptance or rejection of unknown aggregates.
Specifically, there is no clear reason why soundness as tested by ASTM C 88-05
should be a measure of the performance of the given aggregate in concrete subjected
to freezing and thawing.
Other tests consist of subjecting the aggregate to cycles of alternating freezing and
thawing, and sometimes this treatment is applied to mortar or concrete made with the
suspect aggregate. Unfortunately, none of the tests gives an accurate indication of the
behaviour of aggregate under actual conditions of moisture and temperature changes
above the freezing point.
Likewise, there are no tests which could satisfactorily predict the durability of
aggregate in the concrete under conditions of freezing and thawing. The main reason
for this is that the behaviour of aggregate is affected by the presence of the
surrounding hydrated cement paste, so that only a service record can satisfactorily
prove the durability of aggregate.
Nevertheless, certain aggregates are known to be susceptible to frost damage and it
is on these that our attention is centred. These are: porous cherts, shales, some
limestones, particularly laminated limestones, and some sandstones. A common
characteristic of these rocks with a poor record is their high absorption (see Fig.
3.8),3.37 but it should be emphasized that many durable rocks also exhibit high
absorption.

151
 Fig. 3.8. Distribution of sound and unsound aggregate samples as a function of
absorption3.37
For frost damage to occur, there must exist critical conditions of water content and
lack of drainage. These are governed, inter alia, by the size, shape and continuity of
pores in the aggregate because these characteristics of the pores control the rate and
amount of absorption and also the rate at which water can escape from the aggregate
particle. Indeed, these features of the pores are more important than merely their total
volume as reflected by the magnitude of absorption.
It has been found that pores smaller than 4 to 5 μm are critical, because they are
large enough to permit water to enter but not large enough to allow easy drainage
under the pressure of ice. This pressure, in fully confined space at –20 °C (–4 °F), can
be as high as 200 MPa (29 000 psi). Thus, if splitting of aggregate particles and
disruption of the surrounding cement paste are to be avoided, flow of water towards
unfilled pores within the aggregate particle, or into the surrounding paste, must be
possible before the hydraulic pressure becomes high enough to cause disruption.
This argument illustrates the statement made earlier that the durability of aggregate
cannot be fully determined other than when it is embedded in hydrated cement paste:
the particle may be strong enough to resist the pressure of ice but expansion can cause
disruption of the surrounding mortar.
It has been said that the pore size is an important factor in the durability of
aggregate. In most aggregates, pores of different sizes are present so that we are really
confronted with a pore size distribution. A means of expressing this quantitatively has
been developed by Brunauer, Emmett and Teller.3.13 The specific surface of the
aggregate is determined from the amount of a gas sorbate required to form a layer one
molecule thick over the entire internal surface of the aggregate pores. The total
volume of the pores is measured by absorption, and the ratio of the volume of pores to
their surface represents the hydraulic radius of the pores. This value, familiar from
flow problems in hydraulics, gives an indication of the pressure required to produce
flow.
Alkali–silica reaction
In the latter part of the 20th century, an increasing number of deleterious chemical
reactions between the aggregate and the surrounding hydrated cement paste has been

152
observed. The most common reaction is that between the active silica constituents of
the aggregate and the alkalis in cement. The reactive forms of silica are opal
(amorphous), chalcedony (cryptocrystalline fibrous), and tridymite (crystalline).
These reactive materials occur in: opaline or chalcedonic cherts, siliceous limestones,
rhyolites and rhyolitic tuffs, dacite and dacite tuffs, andesite and andesite tuffs, and
phyllites.3.29
The reaction starts with the attack on the siliceous minerals in the aggregate by the
alkaline hydroxides in pore water derived from the alkalis (Na2O and K2O) in the
cement. As a result, an alkali-silicate gel is formed, either in planes of weakness or
pores in the aggregate (where reactive silica is present) or on the surface of the
aggregate particles. In the latter case, a characteristic altered surface zone is formed.
This may destroy the bond between the aggregate and the surrounding hydrated
cement paste.
The gel is of the ‘unlimited swelling’ type: it imbibes water with a consequent
tendency to increase in volume. Because the gel is confined by the surrounding
hydrated cement paste, internal pressures result and may eventually lead to expansion,
cracking and disruption of the hydrated cement paste. Thus, expansion appears to be
due to hydraulic pressure generated through osmosis, but expansion can also be
caused by the swelling pressure of the still solid products of the alkali-silica
reaction.3.30 For this reason, it is believed that it is the swelling of the hard aggregate
particles that is most harmful to concrete. Some of the relatively soft gel is later
leached out by water and deposited in the cracks already formed by the swelling of
the aggregate. The size of the siliceous particles affects the speed with which reaction
occurs, fine particles (20 to 30 μm) leading to expansion within a month or two, larger
ones only after many years.3.60
Reviews of the mechanisms involved in the alkali-silica reactions have been
presented by Diamond3.66 and by Helmuth.3.78 It is believed that the gel formation takes
place only in the presence of Ca++ ions.3.73 This is of importance with reference to the
prevention of expansive aggregate–silica reactions by the inclusion in the mix of
pozzolanas, which remove Ca(OH)2 (see p. 522). The progress of reactions is complex,
but it is important to bear in mind that it is not the presence of the alkali–silica gel per
se but the physico-chemical response to the reactions that leads to the cracking of
concrete.3.66
The alkali–silica reaction occurs only in the presence of water. The minimum
relative humidity in the interior of concrete for the reaction to proceed is about 85 per
cent at 20 °C (68 °F).3.79 At higher temperatures, the reaction can take place at a
somewhat lower relative humidity.3.79 Generally, a higher temperature accelerates the
progress of the alkali–silica reaction but does not increase the total expansion induced
by the reaction.3.79 The effect of temperature may be due to the fact that an increase in
temperature lowers the solubility of Ca(OH)2 and increases that of silica. The
accelerating effect of temperature is exploited in tests on the reactivity of aggregate.
Because water is essential for the alkali–silica reaction to continue, drying out the
concrete and prevention of future contact with water is an effective means of stopping
the reaction; it is, in fact, the only means. Conversely, alternating wetting and drying
aggravates the migration of the alkali ions, which move from the wet to the drier part
of the concrete. A moisture gradient has a similar effect.3.80

153
 The alkali–silica reaction is very slow, and its consequences may not manifest
themselves until after many years. The reasons for this are complex, and the processes
involved, related to the local concentration of various ions, are still debated.3.66
While we can predict that, with given materials, alkali–aggregate reaction will take
place, it is not generally possible to estimate the deleterious effects from the
knowledge of the quantities of the reactive materials alone. For instance, the actual
reactivity of aggregate is affected by its particle size and porosity, which influence the
area over which the reaction can take place. When the alkalis originate from cement
only, their concentration at the reactive surface of aggregate will be governed by the
magnitude of this surface. Within limits, the expansion of concrete made with a given
reactive aggregate is greater the higher the alkali content of the cement and, for a
given alkali content in the cement, the greater its fineness.3.32
Other factors influencing the progress of alkali–aggregate reaction include the
permeability of the hydrated cement paste because this controls the movement of
water and of the various ions, as well as of the silica gel. It can thus be seen that
various physical and chemical factors make the problem of akali–aggregate reaction
highly complex. In particular, the gel can change its constitution by absorption and
thus exert a considerable pressure whereas, at other times, diffusion of the gel out of
the confined area takes place.3.32 It may be noted that, as the hydration of cement
progresses, much of the alkalis becomes concentrated in the aqueous phase. As a
consequence, pH rises and all silica minerals become soluble.3.61
Tests for aggregate reactivity
The preceding discussion explains why it is that, although we know that certain types
of aggregate tend to be reactive, and their presence can be established by ASTM C
295-08, there is no simple way of determining whether a given aggregate will cause
excessive expansion due to reaction with alkalis in the cement. Service record has
generally to be relied upon, but as little as 0.5 per cent of harmful aggregate can cause
damage.3.61 If no record is available, it is possible only to determine the potential
reactivity of the aggregate but not to prove that a deleterious reaction will take place.
A quick chemical test is prescribed by ASTM C 289-07: the reduction in the alkalinity
of a normal solution of NaOH when placed in contact with pulverized aggregate at
80 °C (176 °F) is determined, and the amount of dissolved silica is measured. The
interpretation of the result is in many cases not clear but, generally, a potentially
deleterious reaction is indicated if the plotted test results falls to the right of the
boundary line in Fig. 3.9 reproduced from ASTM C 289-07 but based on Mielenz and
Witte’s paper.3.33 However, potentially deleterious aggregates represented by points
lying above the dashed line in Fig. 3.9 may be extremely reactive with alkalis so that a
relatively low expansion may result. These aggregates should therefore be tested
further to determine whether their reactivity is deleterious by the mortar-bar test
described below. The test is of little value with lightweight aggregates.3.68

154
 Fig. 3.9. Results of chemical test of ASTM C 289-07
In the mortar-bar test for the physical reactivity of aggregate, prescribed by ASTM
C 227-10, the suspected aggregate, crushed if need be and made up to a prescribed
grading, is used in making special cement–sand mortar bars, using a cement with an
equivalent alkali content of more than 0.6 per cent and preferably more than 0.8 per
cent. The bars are stored over water at 38 °C (100 °F), at which temperature the
expansion is more rapid and usually higher than at higher or lower

155
temperatures.3.34 The reaction is also accelerated by the use of a fairly high
water/cement ratio. According to an appendix to ASTM C 33-08, the aggregate under
test is considered harmful if it expands more than 0.10 per cent after 6 months or
more than 0.05 per cent after 3 months, if a 6-month result is not available.
The mortar-bar test of ASTM C 227-10 has shown a very good correlation with
field experience, but a considerable time is required before judgement on the
aggregate can be made. For quartz-bearing aggregates, the test period may need to be
as long as one year.3.81 On the other hand, as mentioned earlier, the results of the
chemical test, which is rapid, are often not conclusive. Likewise, petrographic
examination, although a useful tool in identifying the mineral constituents, cannot
establish that a given mineral will result in deleterious expansion. Various accelerated
test methods continue to be developed but they often involve the use of a high
temperature (up to 80 °C (176 °F)) which distorts the behaviour. The British Standard
prescribing a test method using expansion of a concrete prism is BS 812-123 : 1999.
There is a lack of results of laboratory tests which are correlated with the field
performance of concrete containing the same materials.3.82 The probable reason for this
is the very long period in service before the effects of alkali–silica reaction manifest
themselves. It follows that new test methods cannot be validated in a hurry. A rapid
and conclusive test for aggregate reactivity is thus still to be developed, and to use
more than one of the existing tests is the best that can be done at the moment.
The preceding discussion of the alkali–aggregate reaction is intended to ensure an
awareness of the potential problems with some aggregates. The consequences of the
alkali–aggregate reaction in concrete and the means of avoiding them are discussed on
p. 519, but full treatment of this vast subject cannot be included in the present book.
What is important is to realize that the risk of deleterious alkali–aggregate reaction
has to be considered in the selection of concrete materials.
Alkali–carbonate reaction
Another type of deleterious aggregate reaction is that between some dolomitic
limestone aggregates and the alkalis in cement. The volume of the products of this
reaction is smaller than the volume of the original materials so that the explanation for
the deleterious reaction has to be sought in phenomena different from those involved
in the alkali–silica reaction.3.83 It is likely that the gel which is formed is subject to
swelling in a manner similar to swelling clays.3.79 Thus, under humid conditions,
expansion of concrete takes place. Typically, reaction zones up to 2 mm (or 0.1 in.)
are formed around the active aggregate particles. Cracking develops within these rims
and leads to a network of cracks and a loss of bond between the aggregate and the
cement paste.
Tests have shown that de-dolomitization, that is a change of dolomite, CaMg(CO3)2,
into CaCO3 and Mg(OH)2, occurs. However, the reactions involved are still
imperfectly understood; in particular, the role of clay in the aggregate is not clear, but
expansive reaction seems to be nearly always associated with the presence of clay.
Also, in expansive aggregates the dolomite and calcite crystals are very fine.3.47 One
suggestion is that the expansion is due to moisture uptake by the previously unwetted
clay, the de-dolomitization being necessary only to provide access of moisture to the
locked-in clay;3.48 another suggestion is that clay increases the reactivity of the
aggregate so that dolomite and calcium silicate hydrate produce Mg(OH)2, silica gel,
and calcium carbonate with a volume increase of about 4 per cent.3.62 A good review of
the problem is given by Walker.3.70

156
 It should be stressed that only some dolomitic limestones cause expansive
reactions in concrete. No simple test to identify them has been developed; in case of
doubt, help can be obtained from an investigation of the rock texture or of rock
expansion in sodium hydroxide (ASTM C 586-05). If the expansion of the rock
sample in the test of ASTM C 586-05 exceeds 0.10 per cent, the length change of
concrete made with the suspect aggregate and stored in moist air is determined
according to ASTM C 1105-08a, which also gives guidance on the interpretation of
the test results.
One distinction between the silica- and carbonate–alkali reactions which should be
borne in mind is that, in the latter, the alkali is regenerated. It is probably for this
reason that pozzolanas, including silica fume, are not effective in controlling the
alkali–carbonate expansion.3.84 However, ground granulated blastfurnace slag, which
reduces the permeability of concrete (see Chapter 13) is reasonably
effective.3.84 Fortunately, reactive carbonate rocks are not very widespread and can
usually be avoided.
Thermal properties of aggregate
There are three thermal properties of aggregate that may be significant in the
performance of concrete: coefficient of thermal expansion, specific heat, and
conductivity. The last two are of importance in mass concrete or where insulation is
required, but not in ordinary structural work, and are discussed in the section dealing
with the thermal properties of concrete (see p. 376).
The coefficient of thermal expansion of aggregate influences the value of such a
coefficient of concrete containing the given aggregate: the higher the coefficient of
the aggregate the higher the coefficient of the concrete, but the latter depends also on
the aggregate content in the mix and on the mix proportions in general.
There is, however, another aspect of the problem. It has been suggested that if the
coefficients of thermal expansion of the coarse aggregate and of the hydrated cement
paste differ too much, a large change in temperature may introduce differential
movement and a break in the bond between the aggregate particles and the
surrounding paste. However, possibly because the differential movement is affected
also by other forces, such as those due to shrinkage, a large difference between the
coefficients is not necessarily detrimental when the temperature does not vary outside
the range of, say, 4 to 60 °C (40 to 140 °F). Nevertheless, when the two coefficients
differ by more than 5.5 × 10–6 per °C (3 × 10–6 per °F) the durability of concrete
subjected to freezing and thawing may be affected.
The coefficient of thermal expansion can be determined by means of a dilatometer
devised by Verbeck and Hass3.14 for use with both fine and coarse aggregate. The
linear coefficient of thermal expansion varies with the type of parent rock, the range
for the more common rocks being about 0.9 × 10–6 to 16 × 10–6 per °C (0.5 × 10–6 to 8.9
× 10–6 per °F), but the majority of aggregates lie between approximately 5 × 10–6 and
13 × 10–6 per °C (3 × 10–6 and 7 × 10–6 per °F) (see Table 3.13).3.39 For hydrated Portland
cement paste, the coefficient varies between 11 × 10–6 and 16 × 10–6 per °C (6 × 10–
6 and 9 × 10–6 per °F), but values up to 20.7 × 10–6 per °C (11.5 × 10–6 per °F) have also

been reported, the coefficient varying with the degree of saturation. Thus, a serious
difference in coefficients occurs only with aggregates of a very low expansion; these
are certain granites, limestones and marbles.

Table 3.13. Linear Coefficient of Thermal Expansion of Different Rock Types3.39

157
 If extreme temperatures are expected, the detailed properties of any given
aggregate have to be known. For instance, quartz undergoes inversion at 574 °C and
expands suddenly by 0.85 per cent. This would disrupt the concrete, and for this
reason fire-resistant concrete is never made with quartz aggregate.
Sieve analysis
This somewhat grandiose name is given to the simple operation of dividing a sample
of aggregate into fractions, each consisting of particles of the same size. In practice,
each fraction contains particles between specific limits, these being the openings of
standard test sieves.
The test sieves used for concrete aggregate have square openings and their
properties are prescribed by BS 410-1 and 2 : 2000 and ASTM E 11-09. In the latter
standard, the sieves can be described by the size of the opening (in inches) for larger
sizes, and by the number of openings per lineal inch for sieves smaller than about in.
Thus a No. 100 test sieve has 100 × 100 openings in each square inch. The standard
approach is to designate the sieve sizes by the nominal aperture size in millimetres or
micrometres.
Sieves smaller than 4 mm (0.16 in.) are normally made of wire cloth although, if
required, this can be used up to 16 mm (0.62 in.). The wire cloth is made of phosphor
bronze but, for some coarser sieves, brass and mild steel can also be used. The
screening area, i.e. the area of the openings as a percentage of the gross area of the
sieve, varies between 28 and 56 per cent, being larger for large openings. Coarse test
sieves (4 mm (0.16 in.) and larger) are made of perforated plate, with a screening area
of 44 to 65 per cent.
All sieves are mounted in frames which can nest. It is thus possible to place the
sieves one above the other in order of size with the largest sieve at the top, and the
material retained on each sieve after shaking represents the fraction of aggregate
coarser than the sieve in question but finer than the sieve above. Frames 200 mm (8
in.) in diameter are used for 5 mm ( in.) or smaller sizes, and 300 or 400 mm (12 or
18 in.) diameter frames for 5 mm ( in.) and larger sizes. It may be remembered that
5 mm (or in., No. 4 ASTM) or 4 mm is the dividing line between the fine and
coarse aggregates.

158
 The sieves used for concrete aggregate consist of a series in which the clear
opening of any sieve is approximately one-half of the opening of the next larger sieve
size. The BS test sieve sizes in Imperial units for this series were as follows: 3

in., in., in., in., in., Nos. 7, 14, 25, 52, 100, and 200, and results of tests on
those sieves are still used. Table 3.14 gives the traditional sieve sizes according to
their fundamental description by aperture in millimetres or micrometres and also the
previous British and ASTM designations and approximate apertures in inches.

Table 3.14. Traditional American and British Sieve Sizes

159
 For determination of oversize and undersize aggregate, and especially for research
work on aggregate grading, additional sieve sizes are required. The full sequence of
test sieves is based theoretically on the ratio of for the openings of two
consecutive sieves, 1 mm size being the base. However, both the British (BS 410 :
1986) and American (ASTM E 11-09) sieves have been standardized generally in
accordance with the R40/3 sieve series of the International Standards Organization.
Not all of these sizes form a true geometric series but follow ‘preferred numbers’.
British Standard BS 410-1 : 2000 also uses some sieve sizes of the R20 series of the
International Standards Organization (ISO 565 – 1990). This series covers the range
of sizes from 125 mm to 63 μm in steps with a ratio of approximately 1.2, based on
the 1 mm size. There exists also a European Standard BS EN 933-2 : 1996, which
uses the same sizes as ISO 6274 – 1982. The various standard sieve sizes are shown
in Table 3.15. For grading purposes, the sieve sizes normally used are: 75.0, 50.0,
37.5, 20.0, 10.0, 5.00, 2.36, 1.18 mm, and 600, 300, and 150 μm.

Table 3.15. Standard Sieve Sizes for Aggregate Given by Various Standards (mm
or μm)

160
161
 We can thus see that in discussing aggregate grading we have to contend with two
sets of sieve sizes. In this book, results of measurements made with Imperial size
sieves will be reported by the exact metric equivalent, but grading curves for mix
proportioning purposes (see Chapter 14) will, wherever available, be based on current
ASTM or BS metric sieve sizes.
Before the sieve analysis is performed, the aggregate sample has to be air-dried in
order to avoid lumps of fine particles being classified as large particles and also to
prevent clogging of the finer sieves. The minimum masses of the reduced samples for
sieving, as recommended by BS 812-103.1 : 1985 (2000), are given in Table 3.16,
and Table 3.17 shows the maximum mass of material with which each sieve can cope.
If this mass is exceeded on a sieve, material which is really finer than this sieve may
be included in the portion retained. The material on the sieve in question should,
therefore, be split into two parts and each should be sieved separately. The actual
sieving operation can be performed by hand, each sieve in turn being shaken until not
more than a trace continues to pass. The movement should be backwards and
forwards, sideways left and right, circular clockwise and anticlockwise, all these
motions following one another so that every particle ‘has a chance’ of passing through
the sieve. In most laboratories a sieve shaker is available, usually fitted with a time
switch so that uniformity of the sieving operation can be ensured. None the less, care
is necessary in order to make sure that no sieve is overloaded (see Table 3.17). The
amount of material smaller than 75 μm can best be determined by wet sieving in
accordance with BS 812-103.1 : 1985 (2000) or ASTM C 117-04.

Table 3.16. Minimum Mass of Sample for Sieve Analysis According to BS 812-
103.1 : 1985 (2000)

162
 Table 3.17. Maximum Mass to be Retained at the Completion of Sieving
According to BS 812-103.1 : 1985 (2000)

The results of a sieve analysis are best reported in tabular form, as shown in Table
3.18. Column (2) shows the mass retained on each sieve. This is expressed as a
percentage of the total mass of the sample and is shown in column (3). Now, working
from the finest size upwards, the cumulative percentage (to the nearest 1 per cent)

163
passing each sieve can be calculated (column (4)), and it is this percentage that is used
in the plotting of grading curves.

Table 3.18. Example of Sieve Analysis

Grading curves
The results of a sieve analysis can be grasped much more easily if represented
graphically and, for this reason, grading charts are very extensively used. By using a
chart, it is possible to see at a glance whether the grading of a given sample conforms
to that specified, or is too coarse or too fine, or deficient in a particular size.
In the grading chart commonly used, the ordinates represent the cumulative
percentage passing and the abscissae show the sieve opening plotted to a logarithmic
scale. Since the openings of sieves in a standard series are in the ratio of , a
logarithmic plot shows these openings at a constant spacing. This is illustrated in Fig.
3.10 which represents the data of Table 3.18.

164
 Fig. 3.10. Example of a grading curve (see Table 3.18)
It is convenient to choose a scale such that the scale spacing between two adjacent
sieve sizes is approximately equal to the 20 per cent interval on the ordinate scale; a
visual comparison of different grading curves can then be made from memory.
Fineness modulus
A single factor computed from the sieve analysis is sometimes used, particularly in
the United States. This is the fineness modulus, defined as of the sum of the
cumulative percentages retained on the sieves of the standard series: 150, 300, 600 μm,
1.18, 2.36, 5.00 mm (ASTM Nos. 100, 50, 30, 16, 8, 4) and up to the largest sieve size
used. It should be remembered that, when all the particles in a sample are coarser than,
say, 600 μm (No. 30 ASTM), the cumulative percentage retained on 300 μm (No. 50
ASTM) sieve should be entered as 100; the same value, of course, would be entered
for 150 μm (No. 100). The value of the fineness modulus is higher the coarser the
aggregate (see column (5), Table 3.18).
The fineness modulus can be looked upon as a weighted average size of a sieve on
which the material is retained, the sieves being counted from the finest.
Popovics3.49 showed it to be a logarithmic average of the particle size distribution. For
instance, a fineness modulus of 4.00 can be interpreted to mean that the fourth sieve,
1.18 mm (No. 16 ASTM) is the average size. However, it is clear that one parameter,
the average, cannot be representative of a distribution: thus the same fineness
modulus can represent an infinite number of totally different size distributions or
grading curves. The fineness modulus cannot, therefore, be used as a single
description of the grading of an aggregate, but it is valuable for measuring slight
variations in the aggregate from the same source, e.g. as a day-to-day check.
Nevertheless, within certain limitations, the fineness modulus gives an indication of

165
the probable behaviour of a concrete mix made with aggregate having a certain
grading, and the use of the fineness modulus in assessment of aggregates and in mix
proportioning has many supporters.3.49
Grading requirements
We have seen how to find the grading of a sample of aggregate, but it still remains to
determine whether or not a particular grading is suitable. A related problem is that of
combining fine and coarse aggregates so as to produce a desired grading. What, then,
are the properties of a ‘good’ grading curve?
Because the strength of fully compacted concrete with a given water/cement ratio
is independent of the grading of the aggregate, grading is, in the first instance, of
importance only in so far as it affects workability. As, however, achieving the strength
corresponding to a given water/cement ratio requires full compaction, and this can be
obtained only with a sufficiently workable mix, it is necessary to produce a mix that
can be compacted to a maximum density with a reasonable amount of work.
It should be stated at the outset that there is no one ideal grading curve but a
compromise is aimed at. Apart from the physical requirements, the economic aspects
must not be forgotten: concrete has to be made of materials which can be produced
cheaply so that no narrow limits can be imposed on aggregate.
It has been suggested that the main factors governing the desired aggregate grading
are: the surface area of the aggregate, which determines the amount of water
necessary to wet all the solids; the relative volume occupied by the aggregate; the
workability of the mix; and the tendency to segregation.
Segregation is discussed on p. 205, but it should be observed here that the
requirements of workability and absence of segregation tend to be partially opposed to
one another: the easier it is for the particles of different sizes to pack, smaller particles
passing into the voids between the larger ones, the easier it is also for the small
particles to be shaken out of the voids, i.e. to segregate in the dry state. In actual fact,
it is the mortar (i.e. a mixture of sand, cement and water) that should be prevented
from passing freely out of the voids in the coarse aggregate. It is also essential for the
voids in the combined aggregate to be sufficiently small to prevent the fresh cement
paste from passing through and separating out.
The problem of segregation is thus rather similar to that of filters, although the
requirements in the two cases are of course diametrically opposite: for the concrete to
be satisfactory it is essential that segregation be avoided.
There is a further requirement for a mix to be satisfactorily cohesive and workable:
it must contain a sufficient amount of material smaller than a 300 μm (No. 50 ASTM)
sieve. Because the cement particles are included in this material, a richer mix requires
a lower content of fine aggregate than a lean mix. If the grading of fine aggregate is
such that it is deficient in finer particles, increasing the fine/coarse aggregate ratio
may not prove a satisfactory remedy, as it may lead to an excess of middle sizes and
possibly to harshness. (A mix is said to be harsh when one size fraction is present in
excess, as shown by a steep step in the middle of a grading curve, so that particle
interference results.) This need for an adequate amount of fines (provided they are
structurally sound) explains why minimum contents of particles passing 300 μm (No.
50 ASTM) and sometimes also 150 μm (No. 100) sieves are laid down, as for instance
in Tables 3.22 and 3.23 (p. 167). However, it is now thought that the U.S. Bureau of

166
Reclamation requirements of Table 3.23 for the minimum percentage of particles
passing the 300 and 150 μm (Nos 50 and 100 ASTM) sieves are too high.
It may be further added that all the cementitious materials automatically provide a
certain amount of ‘ultra-fines’. The ultra-fines can be, therefore, taken as materials
smaller than 125 μm of all provenance, that is, aggregate, filler, and cement. However,
there are some differences in behaviour in that the initial hydration of cement rapidly
removes some water from the mix, while the other particles are inert. The volume of
entrained air can be taken as equivalent to one-half the volume of fines. The German
Standard DIN 1045 : 19883.86 establishes the particle size of 125 μm as the criterion for
ultra-fine material. No minima of ultra-fines are specified because they are normally
found in the materials used, but the presence of adequate ultra-fines is essential for
pumped concrete and for concrete to be placed in thin sections or with congested
reinforcement, and also for water-retaining structures. On the other hand, an excessive
amount of ultra-fines is harmful from the point of view of resistance to freezing and
thawing and to de-icing salts as well as of resistance to abrasion. A maximum total
content of 350 kg per cubic metre of concrete is prescribed for mixes with a cement
content of not more than 300 kg/m3. The maximum of ultra-fines is 400 kg/m3 when
the cement content is 350 kg/m3; higher amounts of ultra-fines are permitted at higher
cement contents. These values apply to mixes with a maximum aggregate size of 16
to 63 mm. The beneficial effect of ultra-fines smaller than 50 μm on the water
requirement of fresh concrete, and therefore on strength, has been confirmed.3.85
The requirement that the aggregate occupies as large a relative volume as possible
is, in the first instance, an economic one, the aggregate being cheaper than the cement
paste, but there are also strong technical reasons why too rich a mix is undesirable. It
is also believed that the greater the amount of solid particles that can be packed into a
given volume of concrete the higher its density and therefore the higher its strength.
This maximum density theory has led to the advocacy of grading curves parabolic in
shape, or in part parabolic and then straight (when plotted to a natural scale), as
shown in Fig. 3.11. It was found, however, that the aggregate graded to give
maximum density makes a harsh and somewhat unworkable mix. The workability is
improved when there is an excess of paste above that required to fill the voids in the
sand, and also an excess of mortar (fine aggregate plus cement paste) above that
required to fill the voids in the coarse aggregate.

167
 Fig. 3.11. Fuller’s grading curves
The concept of an ‘ideal’ grading curve, such as that shown in Fig. 3.11, still finds
favour, although somewhat varying shapes of ‘ideal’ curves are recommended by
different researchers.3.87
One ‘ideal’ grading derived from the asphalt industry, in which it is important to
minimize the volume of the binder, is as follows. A graph is constructed in which the
ordinate is the cumulative percentage passing and the abscissa represents the sieve
size raised to the power of 0.45. A straight line is drawn on this graph connecting a
point corresponding to the largest sieve size on which some aggregate is retained, to a
point corresponding to the sieve size onto which no more aggregate arrives in sieving.
The ‘ideal’ grading should follow this line, except that the percentage passing from
600 μm (No. 30 ASTM) sieve downwards should fall below the straight line, which
does not take into account the presence of cement – also a fine material. It is claimed
that gradings which do not swing widely above and below the straight line produce
dense concrete, but the ‘0.45 power grading curve’ approach is not proven and not
widely used.
The practical problem is that aggregates from different sources, even if nominally
of the same grading, vary in the actual distribution of particle size within a given size
fraction, as well as in other properties of the particles such as shape and texture. It has
to be added that the total volume of voids in concrete is reduced when the range of
particle sizes from the maximum aggregate size downward is as large as possible, that
is, if extremely fine particles are included in the mix; silica fume, which is one such
material, is considered on p. 86.
Let us now consider the surface area of the aggregate particles. The water/cement
ratio of the mix is generally fixed from strength considerations. At the same time, the
amount of the fresh cement paste has to be sufficient to cover the surface of all the
particles so that the lower the surface area of the aggregate the less paste, and
therefore the less water, is required.

168
 Taking for simplicity a sphere of diameter D as representative of the shape of the
aggregate, we have the ratio of the surface area to volume of 6/D. This ratio of the
surface of the particles to their volume (or, when the particles have a constant specific
gravity, to their mass) is called specific surface. For particles of a different shape, a
coefficient other than 6/D would be obtained but the surface area is still inversely
proportional to the particle size, as shown in Fig. 3.12 reproduced from Shacklock and
Walker’s report.3.15 It should be noted that a logarithmic scale is used for both the
ordinates and the abscissae because the sieve sizes are in geometrical progression.

Fig. 3.12. Relation between specific surface and particle size3.15

In the case of graded aggregate, the grading and the overall specific surface are
related to one another, although of course there are many grading curves
corresponding to the same specific surface. If the grading extends to a larger
maximum aggregate size, the overall specific surface is reduced and the water
requirement decreases, but the relation is not linear. For instance, increasing the

maximum aggregate size from 10 mm to 63 mm ( in. to in.) can, under certain

conditions, reduce the water requirement for a constant workability by as much as 50
kg per cubic metre of concrete (85 lb/yd3). The corresponding decrease in the
water/cement ratio may be as much as 0.15.3.16 Some typical values are shown in Fig.
3.13.

169
Fig. 3.13. Influence of maximum size of aggregate on mixing water requirement
for a constant slump3.16
The practical limitations of the maximum size of aggregate that can be used under
given circumstances and the problem of influence of the maximum size on strength in
general are discussed on p. 174.
It can be seen that, having chosen the maximum size of aggregate and its grading,
we can express the total surface area of the particles using the specific surface as a
parameter, and it is the total surface of the aggregate that determines the water
requirement or the workability of the mix. Mix design on the basis of the specific
surface of the aggregate was first suggested by Edwards3.50 as far back as 1918, and
interest in this method was renewed 40 years later. Specific surface can be determined
using the water permeability method,3.17 but no simple field test is available, and a
mathematical approach is made difficult by the variability in the shape of different
aggregate particles.
This, however, is not the only reason why the selection of mix proportions on the
basis of the specific surface of aggregate is not universally recommended. The
application of surface area calculations was found to break down for aggregate
particles smaller than about 150 μm (No. 100 ASTM) sieve, and for cement. These
particles, and also some larger sand particles, appear to act as a lubricant in the mix
and do not seem to require wetting in quite the same way as coarse particles. An
indication of this was found in some tests by Glanville et al.3.18
Because specific surface gives a somewhat misleading picture of the workability to
be expected (largely owing to an overestimate of the effect of fine particles), an

170
empirical surface index was suggested by Murdock3.19 and its values as well as those of
the specific surface are given in Table 3.19.

Table 3.19. Relative Values of Surface Area and Surface Index

The overall effect of the surface area of an aggregate of given grading is obtained
by multiplying the percentage mass of any size fraction by the coefficient
corresponding to that fraction, and summing all the products. According to
Murdock,3.19 the surface index (modified by an angularity index) should be used, and
in fact the values of this index are based on empirical results. On the other hand,
Davey3.20 found that, for the same total specific surface of the aggregate, the water
requirement and the compressive strength of the concrete are the same for very wide
limits of aggregate grading. This applies both to continuously- and gap-graded
aggregate, and in fact three of the four gradings listed in Table 3.20, reproduced from
Davey’s paper, are of the gap type.

Table 3.20. Properties of Concretes Made with Aggregates of the Same Specific
Surface3.20

171
 An increase in the specific surface of the aggregate for a constant water/cement
ratio has been found to lead to a lower strength of concrete, as shown for instance
in Table 3.21, showing Newman and Teychenné’s3.21 results. The reasons for this are
not quite clear, but it is possible that a reduction in density of the concrete consequent
upon an increase in the fineness of the natural sand is instrumental in lowering the
strength.3.22

Table 3.21. Specific Surface of Aggregate and Strength of Concrete for a 1:6 Mix
with a Water/Cement Ratio of 0.603.21

Workability does not seem to be a direct function of the specific surface area of
aggregate; indeed, Hobbs3.88 showed that concrete mixes containing fine aggregate
with significantly varying grading led to a similar slump or compacting factor, but the
percentage of fine aggregate in the total aggregate was adjusted. It seems then that the
surface area of the aggregate is an important factor in determining the workability of
the mix, but the exact role played by the finer particles has by no means been
ascertained.
The type gradings of Road Note No. 43.23 which is an early fundamental
contribution to understanding aggregate grading, represent different values of overall
specific surface. For instance, when river sand and gravel are used, the four grading
curves, Nos 1 to 4, of Fig. 3.14 correspond to the specific surface of 1.6, 2.0, 2.5, and
3.3 m2/kg, respectively.3.21 In practice, when trying to approximate type gradings, the
properties of the mix will remain largely unaltered when compensation of a small
deficiency of fines by a somewhat larger excess of coarser particles is applied, but the
departure must not be too great. The deficiency and excess are, of course, mutually
interchangeable in the above statement.

172
 Fig. 3.14. Road Note No. 4 type grading curves for 19.05 mm ( in.)
aggregate3.23 (Crown copyright)
There is no doubt then that the grading of aggregate is a major factor in the
workability of a concrete mix. Workability, in turn, affects the water and cement
requirements, controls segregation, has some effect on bleeding, and influences the
placing and finishing of the concrete. These factors represent the important
characteristics of fresh concrete and affect also its properties in the hardened state:
strength, shrinkage, and durability.
Grading is thus of vital importance in the proportioning of concrete mixes, but its
exact role in mathematical terms has not yet been established, and the behaviour of
this type of semi-liquid mixture of granular materials is still imperfectly understood.
Moreover, while ensuring appropriate grading of aggregate is of considerable
importance, arbitrary imposition of limits which are uneconomic, or even near
impossible, in a given location is inappropriate.
Finally, it must be remembered that far more important than devising a ‘good’
grading is ensuring that the grading is kept constant; otherwise, variable workability
results and, when this is corrected at the mixer by a variation in the water content,
concrete of variable strength is obtained.
Practical gradings
From the brief review in the previous section, it can be seen how important it is to use
aggregate with a grading such that a reasonable workability and minimum segregation
are obtained. The importance of the latter requirement cannot be over-emphasized: a
workable mixture which could produce a strong and economical concrete will result

173
in honeycombed, weak, not durable and variable end product if segregation takes
place.
The process of calculation of the proportions of aggregates of different size to
achieve the desired grading comes within the scope of mix proportioning and is
described in Chapter 14. Here, the properties of some ‘good’ grading curves will be
discussed. It should be remembered, however, that in practice the aggregate available
locally or within an economic distance has to be used, and this can generally produce
satisfactory concrete, given an intelligent approach and sufficient care. For aggregate
which includes natural sand, it may be useful, as one basis of comparison, to use the
curves of the Road Research Note No. 4 on the Design of Concrete Mixes.3.23 They
have been prepared for aggregates of 19.05 and 38.1 mm ( in. and in.) maximum
size, and are reproduced in Figs 3.14 and 3.15, respectively. Similar curves for

aggregate with a 9.52 mm ( in.) maximum size have been prepared by McIntosh and
Erntroy,3.24 and are shown in Fig. 3.16.

Fig. 3.15. Road Note No. 4 type grading curves for 38.1 mm ( in.)
aggregate3.23 (Crown copyright)

174
 Fig. 3.16. Mclntosh and Erntroy’s type grading curves for 9.52 mm ( in.)
aggregate3.24
Four curves are shown for each maximum size of aggregate but, due to the
presence of over- and under-size aggregate and also because of variation within any
fraction size, practical gradings are more likely to lie in the vicinity of these curves
than to follow them exactly. It is therefore preferable to consider grading zones, and
these are marked on all the diagrams.
Curve No. 1 represents the coarsest grading in each of the Figs 3.14 to 3.16. Such a
grading is comparatively workable and can, therefore, be used for mixes with a low
water/cement ratio or for rich mixes; it is, however, necessary to make sure that
segregation does not take place. At the other extreme, curve No. 4 represents a fine
grading: it will be cohesive but not very workable. In particular, an excess of material
between 1.20 and 4.76 mm (No. 16 and in.) sieves will produce a harsh concrete,
which may be suitable for compaction by vibration, but is difficult to place by hand. If
the same workability is to be obtained using aggregates with grading curves Nos 1
and 4, the latter would require a considerably higher water content: this would mean a
lower strength if both concretes are to have the same aggregate/cement ratio or, if the
same strength is required, the concrete made with the finer aggregate would have to
be considerably richer, i.e. each cubic metre would contain more cement than when
the coarser grading is used.
The change between the extreme gradings is progressive. In the case of gradings
lying partly in one zone, partly in another, there is, however, a danger of segregation
when too many intermediate sizes are missing (cf. gap grading). If, on the other hand,
there is an excess of middle-sized aggregate, the mix will be harsh and difficult to

175
compact by hand and possibly even by vibration. For this reason, it is preferable to
use aggregate with gradings similar to type, rather than totally dissimilar ones.
Figures 3.17 and 3.18 show the range of gradings used with 152.4 mm (6 in.) and
76.2 mm (3 in.) maximum aggregate size, respectively, as given by McIntosh.3.25 The
actual gradings, as usual, run parallel with the limit curves rather than crossing over
from one to the other.

Fig. 3.17. Range of gradings used with 152.4 mm (6 in.) aggregate3.25

Fig. 3.18. Range of gradings used with 76.2 mm (3 in.) aggregate3.25

176
 In practice, the use of separate fine and coarse aggregate means that a grading can
be made up to conform exactly with a type grading at one intermediate point,
generally the 5 mm ( in.) size. Good agreement can usually also be obtained at the
ends of the curve (150 μm (No. 100) sieve and the maximum size used). If coarse
aggregate is delivered in single-size fractions, as is usually the case, agreement at
additional points above 5 mm ( in.) can be obtained, but for sizes below 5 mm
( in.) blending of two or more fine aggregates is necessary.
Grading of fine and coarse aggregates
Given that, for any but unimportant work, fine and coarse aggregates are batched
separately, the grading of each part of the aggregate should be known and controlled.
Over the years, there have been several approaches to specifying the grading
requirements for fine aggregate. First, type grading curves were given as representing
‘good’ grading.3.23 In the 1973 edition of BS 882, four grading zones were introduced.
The division into zones was based primarily on the percentage passing the 600 μm
(No. 30 ASTM) sieve. The main reason for this was that a large number of natural
sands divide themselves at just that size, the gradings above and below being
approximately uniform. Furthermore, the content of particles finer than the 600 μm
(No. 30 ASTM) sieve has a considerable influence on the workability of the mix and
provides a fairly reliable index of the overall specific surface of the sand.
Thus, the grading zones largely reflected the grading of natural sands available in
the United Kingdom. Little of these sands is now available for concrete-making, and a
much less restrictive approach to grading is reflected in the requirements of BS 882 :
1992. This does not mean that ‘any grading will do’; rather, given that grading is but
one feature of aggregate, a wide range of gradings may be acceptable but a trial-and-
error approach is required.
Specifically, BS 882 : 1992 (withdrawn in 2004) requires any fine aggregate to
satisfy the overall grading limits of Table 3.22 and also one of the three additional
grading limits of the same table, but one in ten consecutive samples is allowed to fall
outside the additional limits. The additional limits are, in effect, a coarse, a medium,
and a fine grading. The current standard is BS EN 12620 : 2002.

Table 3.22. BS and ASTM Grading Requirements for Fine Aggregate

177
 The requirements of BS 882 : 1992 may be inappropriate for some precast concrete
and should not be applied in such cases.
For comparison, the requirements of ASTM C 33-08, are in part, included in Table
3.22. ASTM C 33-08 also requires the fine aggregate to have a fineness modulus of
between 2.3 and 3.1. The requirements of the U.S. Bureau of Reclamation3.74 are given
in Table 3.23. It may be noted that, in the case of air-entrained concrete, lower
quantities of the finest particles are acceptable, the entrained air acting effectively as
very fine aggregate. ASTM C 33-08 also allows reduced percentages passing sieves
300 and 150 μm (Nos 50 and 100 ASTM) when the cement content is above 297
kg/m3 (500 lb/yd3) or if air entrainment is used with at least 237 kg of cement per
cubic metre of concrete (400 lb/yd3).

Table 3.23. U.S. Bureau of Reclamation Grading Requirements for Fine

Aggregate3.74

178
 Fine aggregate satisfying any of the additional grading limits of BS 882 : 1992 can
generally be used in concrete, although under some circumstances the suitability of a
given fine aggregate may depend on the grading and shape of the coarse aggregate.
Crushed fine aggregate tends to have different grading from most natural sands.
Specifically, there is less material between 600 and 300 μm (Nos 30 and 50) sieve
sizes, coupled with more material larger than 1.18 mm (No. 16) sieve size and also
more very fine material, smaller than 150 or 75 μm (No. 100 or No. 200) sieve size.
Most specifications recognize the last feature and allow a higher content of very fine
particles in crushed fine aggregate. It is important to ensure that this very fine material
does not include clay or silt.
It has been shown3.71 that increasing the content of particles smaller than 150 μm
(No. 100) in crushed rock fine aggregate from 10 to 25 per cent results in only a small
decrease in the compressive strength of concrete, typically by 10 per cent.
In considering the effects of a large amount of very fine material in the aggregate,
it is useful to note that, when the material is well-rounded and smooth, workability is
improved, and this is advantageous in terms of reduced water demand. Fine dune
sands have such characteristics.3.38
In general terms, the ratio of coarse to fine aggregate should be higher the finer the
grading of the fine aggregate. When crushed rock coarse aggregate is used, a slightly
higher proportion of fine aggregate is required than with gravel aggregate in order to
compensate for the lowering of workability by the sharp, angular shape of the crushed
particles.
The requirements of BS 882 : 1992 for the grading of coarse aggregate are
reproduced in Table 3.24: values are given both for graded aggregate and for nominal
one-size fractions. The current British Standard BS EN 12620 : 2000 is augmented by
PD 6682-1 : 2009. For comparison, some of the limits of ASTM C 33-08 are given
in Table 3.25.

Table 3.24. Grading Requirements for Coarse Aggregate According to BS 882 :

1992

179
Table 3.25. Grading Requirements for Coarse Aggregate According to ASTM C
33-08

The actual grading requirements depend, to some extent, on the shape and surface
characteristics of the particles. For instance, sharp, angular particles with rough
surfaces should have a slightly finer grading in order to reduce the possibility of
interlocking and to compensate for the high friction between the particles. The actual
grading of crushed aggregate is affected primarily by the type of crushing plant
employed. A roll granulator usually produces fewer fines than other types of crushers,
but the grading depends also on the amount of material fed into the crusher.
The grading limits for all-in aggregate prescribed by BS 882 : 1992 (substantially
retained in PD 6682-1 : 2009) are reproduced in Table 3.26. It should be remembered
that this type of aggregate is not used except for small and unimportant jobs, mainly
because it is difficult to avoid segregation in stockpiling.

Table 3.26. Grading Requirements for All-in Aggregate According to BS 882 :

1992

180
Oversize and undersize
Strict adherence to size limits of aggregate is not possible: breakage during handling
will produce some undersize material, and wear of screens in the quarry or at the
crusher will result in oversize particles being present.
In the United States, it is usual to specify over- and undersize screen sizes
as and , respectively, of the nominal sieve size;3.74 actual values are given in Table
3.27. The quantity of aggregate smaller than the undersize or larger than the oversize
is generally severely limited.

Table 3.27. Sizes of Over- and Under-Size Screens of U.S. Bureau of

Reclamation3.74

181
 The grading requirements of BS 882 : 1992 allow some under- and oversize for
coarse aggregate. The values, given in Table 3.24, show that between 5 and 10 per
cent oversize is permitted. However, no aggregate must be retained on a sieve one
size larger (in the standard series) than the nominal maximum size. In the case of
single-size aggregate, some undersize is also allowed, and the amount passing the
sieve next smaller than the nominal size is also prescribed. It is important that this fine
fraction of coarse aggregate be not neglected in the calculation of the actual grading.
For fine aggregate, BS 882 : 1992 allows 11 per cent oversize (see Table 3.22).
General grading requirements for coarse and fine aggregate are given in BS EN
12620 : 2002 in terms of upper sieve size D and lower sieve size d with D/d ≥ 1.4.
Gap-graded aggregate
As mentioned earlier, aggregate particles of a given size pack so as to form voids that
can be penetrated only if the next smaller size of particles is sufficiently small, that is,
there is no particle interference. This means that there must be a minimum difference
between the sizes of any two adjacent particle fractions. In other words, sizes
differing but little cannot be used side by side, and this has led to advocacy of gap-
graded aggregate.
Gap grading can then be defined as a grading in which one or more intermediate
size fractions are omitted. The term continuously graded is used to describe
conventional grading when it is necessary to distinguish it from gap grading. On a
grading curve, gap grading is represented by a horizontal line over the range of sizes
omitted. For instance, the top grading curve of Fig. 3.19 shows that no particles of

size between 10.0 and 2.36 mm ( in. and No. 8 ASTM) sieve are present. In some

cases, a gap between 10.0 and 1.18 mm ( in. and No. 16 ASTM) sieves is considered
suitable. Omission of these sizes would reduce the number of stockpiles of aggregate
required and lead to economy. In the case of aggregate of 20.0 mm ( in.) maximum

size, there would be two stockpiles only: 20.0 to 10.0 mm ( to in.), and fine
aggregate screened through a 1.18 mm (No. 16 ASTM) screen. The particles smaller
than 1.18 mm (No. 16 ASTM) sieve size could easily enter the voids in the coarse
aggregate so that the workability of the mix would be higher than that of a
continuously graded mix with the same fine aggregate content.

182
 Fig. 3.19. Typical gap gradings
Tests by Shacklock3.26 have shown that, for a given aggregate/cement ratio and
water/cement ratio, a higher workability is obtained with a lower fine aggregate
content in the case of gap-graded aggregate than when continuously graded aggregate
is used. However, in the more workable range of mixes, gap-graded aggregate showed
a greater proneness to segregation. For this reason, gap grading is recommended
mainly for mixes of relatively low workability: such mixes respond well to vibration.
Good control and, above all, care in handling, so as to avoid segregation, are essential.
It may be observed that, even when some ‘ordinary’ aggregates are used, gap
grading exists; for instance, the use of very fine sand, as found in many countries,
means that there is a deficiency of particles between the 5.00 mm and 2.36 or 1.18
mm ( in. and No. 8 or No. 16 ASTM) sieve sizes. Thus, whenever we use such a
sand without blending it with a coarser sand, we are, in fact, using a gap-graded
aggregate.
Gap-graded aggregate concrete is difficult to pump because of the danger of
segregation, and is not suited to slip-form paving. Otherwise, gap-graded aggregate
can be used in any concrete, but there are two cases of interest: preplaced aggregate
concrete (see p. 228) and exposed aggregate concrete; in the latter, a pleasing finish is
obtained because a large quantity of only one size of coarse aggregate becomes
exposed after treatment.

183
 From time to time, various claims of superior properties have been made for
concrete made with gap-graded aggregate, but these do not seem to have been
substantiated. Strength, both compressive and tensile, does not appear to be affected.
Likewise, Fig. 3.20, showing McIntosh’s3.27 results, confirms that, using given
materials with a fixed aggregate/cement ratio (but adjusting the fine aggregate
content), approximately the same workability and strength are obtained with gap and
continuous gradings; Brodda and Weber3.72 reported a slight negative influence of gap
grading on strength.

Fig. 3.20. Workability and strength of 1 : 6 concretes made with gap- and
continuously graded aggregates.3.27 Cross denotes gap-graded and circle
continuously graded mixes. Each group of points represents mixes with the
water/cement ratio indicated but with different sand contents
Similarly, there is no difference in shrinkage of the concretes made with aggregate
of either type of grading,3.26 although it might be expected that a framework of coarse
particles almost touching one another would result in a lower total change in
dimensions on drying. The resistance of concrete to freezing and thawing is lower
when gap-graded aggregate is used.3.26
It seems, therefore, that the rather extravagant claims made by advocates of gap
grading are not borne out. The explanation lies probably in the fact that, while gap
grading makes it possible for maximum packing of particles to occur, there is no way
of ensuring that it will occur. Both gap-graded and continuously graded aggregate can
be used to make good concrete, but in each case the right percentage of fine aggregate

184
has to be chosen. Thus, once again, it can be seen that we should not aim at any ideal
grading but find the best combination of the available aggregates.
Maximum aggregate size
It has been mentioned before that the larger the aggregate particle the smaller the
surface area to be wetted per unit mass. Thus, extending the grading of aggregate to a
larger maximum size lowers the water requirement of the mix, so that, for a specified
workability and cement content, the water/cement ratio can be lowered with a
consequent increase in strength.

This behaviour has been verified by tests with aggregates up to 38.1 mm ( in.)
maximum size,3.28 and is usually assumed to extend to larger sizes as well.
Experimental results show, however, that above the 38.1 mm ( in.) maximum size
the gain in strength due to the reduced water requirement is offset by the detrimental
effects of lower bond area (so that volume changes in the paste cause larger stresses at
interfaces) and of discontinuities introduced by the very large particles, particularly in
rich mixes. Concrete becomes grossly heterogeneous and the resultant lowering of
strength may possibly be similar to that caused by a rise in the crystal size and
coarseness of texture in rocks.
This adverse effect of increase in the size of the largest aggregate particles in the
mix exists, in fact, throughout the range of sizes, but below 38.1 mm ( in.) the
effect of size on the decrease in the water requirement is dominant. For larger sizes,
the balance of the two effects depends on the richness of the mix,3.42,3.51 as shown in Fig.
3.21. Nichols3.89 confirmed that, for any given strength of concrete, that is, for a given
water/cement ratio, there is an optimum maximum size of aggregate.

185
 Fig. 3.21. Influence of maximum size of aggregate on the 28-day compressive
strength of concretes of different richness3.51
Thus, the best maximum size of aggregate from the standpoint of strength is a
function of the richness of the mix. Specifically, in lean concrete (165 kg of cement
per cubic metre (280 lb/yd3)), the use of 150 mm (or 6 in.) aggregate is advantageous.
However, in structural concrete of usual proportions, from the point of view of
strength, there is no advantage in using aggregate with a maximum size greater than
about 25 or 40 mm (1 or in.). Moreover, the use of larger aggregate would require
the handling of a separate stockpile and might increase the risk of segregation,
especially when the maximum size is 150 mm (6 in.). However, a practical decision
would also be influenced by the availability and cost of different size fractions. The
choice of the maximum size of aggregate in high performance concrete is discussed
on p. 678.
There are, of course, structural limitations too: the maximum size of aggregate
should be no more than to of the thickness of the concrete section and is related
also to the spacing of reinforcement. The governing values are prescribed in codes of
practice.
Use of ‘plums’

186
The original idea of the use of aggregate as an inert filler can be extended to the
inclusion of large stones in a normal concrete: thus the apparent yield of concrete for
a given amount of cement is increased. The resulting concrete is sometimes called
cyclopean concrete.
These large stones are called ‘plums’ and, used in a large concrete mass, they can
be as big as a 300 mm (1 ft) cube but should not be greater than one-third of the least
dimension to be concreted. The volume of plums should not exceed 20 to 30 per cent
of the total volume of the finished concrete, and they have to be well dispersed
throughout the mass. This is achieved by placing a layer of normal concrete, then
spreading the plums, followed by another layer of concrete, and so on. Each layer
should be of such thickness as to ensure at least 100 mm (4 in.) of concrete around
each plum. Care must be taken to ensure that no air is trapped underneath the stones
and that the concrete does not work away from their underside. The plums must have
no adhering coating. Otherwise, discontinuities between the plums and the concrete
may induce cracking and adversely affect permeability.
The placing of plums requires a large amount of labour and also breaks the
continuity of concreting. It is, therefore, not surprising that, with the current high ratio
of the cost of labour to the cost of cement, the use of plums is not economical except
under special circumstances.
Handling of aggregate
Handling and stockpiling of coarse aggregate can easily lead to segregation. This is
particularly so when discharging and tipping permits the aggregate to roll down a
slope. A natural case of such segregation is a scree (talus): the size of particles is
uniformly graded from largest at the bottom to smallest at the top.
A description of the precautions necessary in handling operations is outside the
scope of this book, but one vital recommendation should be mentioned: coarse
aggregate should be split into size fractions approximately 5 to 10, 10 to 20, 20 to 40

mm (or to , to , to in.), etc. These fractions should be handled and

stockpiled separately and remixed only when being fed into the concrete mixer in the
desired proportions. Thus, segregation can occur only within the narrow size range of
each fraction, and even this can be reduced by careful handling procedures.
Care is necessary to avoid breakage of the aggregate: particles greater than 40 mm
(or in.) should be lowered into bins by means of rock ladders and not dropped from
a height.
On large and important jobs, the results of segregation and breakage in handling
(i.e. excess of undersize particles) are eliminated by ‘finish rescreening’ immediately
prior to feeding into the batching bins over the mixer. The proportions of different
sizes are thus controlled much more effectively but the complexity and cost of the
operations are correspondingly increased. This is, however, repaid by easier placing
of uniformly workable concrete and by a possible saving in cement due to the
uniformity of the concrete.
Improper handling of aggregates can result in contamination by other aggregate or
by deleterious material: it was observed on one occasion that aggregate was being
transported in sacks which had previously contained sugar (see p. 253).
Special aggregates

187
This chapter has been concerned solely with natural aggregate of normal weight;
lightweight aggregates are discussed in Chapter 13. There exist, however, also other
aggregates of normal weight, or nearly so, which are artificial in origin. The reasons
for their advent on the concrete scene are as follows.
Environmental considerations are increasingly affecting the supply of aggregate.
There are strong objections to opening of pits as well as to quarrying. At the same
time, there are problems with the disposal of construction demolition waste and with
dumping of domestic waste. Both these types of waste can be processed into
aggregate for use in concrete, and this is increasingly being done.
Recycled concrete aggregate
The aggregate obtained by comminution of demolished concrete is known as recycled
concrete aggregate (RCA). Until now, RCA has been used predominantly in highway
pavements and in non-structural concrete. There is, however, little doubt that
structural use of RCA will increase, but care is required. According to ASTM C 294-
05, RCA is classified as an artificial aggregate. The following specific points should
be considered when using old concrete as aggregate for making new concrete.
Because RCA consists in part of old mortar, the unit weight (density) of concrete
made with RCA is lower than that of concrete made with conventional aggregate. For
the same reason, concrete made with RCA has a higher porosity and absorption. The
higher absorption of RCA can be exploited if it is saturated before mixing: the
absorbed water provides internal curing. In particular, this is so for RCA containing a
large amount of brick.
The potential compressive strength of the new concrete is largely controlled by the
strength of the old concrete, provided the fine aggregate is crushed rock or natural
sand of suitable quality. A substantial reduction in compressive strength may result if
conventional fine aggregate is replaced partly or wholly by fine aggregate from the
old concrete; anything smaller than 2 mm should be discarded. The use of RCA
decreases workability of fresh concrete at a given water content, increases water
requirement at a given consistency, increases drying shrinkage at a given water
content, and reduces the modulus of elasticity at a given water/cement ratio. Those
effects are greatest when the old concrete is used as both coarse and fine aggregate.
Freezing and thawing resistance of the new concrete depends on the air-void system
and strength of the old concrete as well as the corresponding properties of the new
concrete.
Chemical, air-entraining and mineral admixtures present in the old concrete will
not significantly modify the properties of the new concrete. However, high
concentrations of chloride ions in the old concrete may contribute to accelerated
corrosion of steel embedments in the new concrete. Prospective sources of old
concrete may be unsuitable if they were subjected to aggressive chemical attack or
leaching, damage by fire or service at high temperature, etc.
The significance of contaminants in the old concrete, such as noxious, toxic or
radioactive substances, should be analysed in relation to the anticipated service of the
new concrete. While the presence of bituminous materials may impair air-entrainment,
appreciable concentrations of organic materials could produce excessive air-
entrainment. Metallic inclusions may cause rust staining or surface blistering, and
glass fragments could lead to alkali-aggregate reaction.
A method for determining the composition of RCA is given in BS 8500-2 : 2002.

188
 The necessary treatment of waste is not simple and the use of aggregate made from
waste requires specialist knowledge as none of the materials has become standardized.
In particular, building rubble may contain deleterious amounts of brick, glass, gypsum
or chlorides.3.31,3.35,3.36 The processing of demolition waste so as to convert it into
satisfactory aggregate free from contaminants is still being developed. A reduction in
aggregate interlock in concrete made with recycled aggregate has been confirmed by
González et al.3.90 The influence of aggregate type on interlock is discussed by
Regan.3.91
As far as the use of domestic refuse is concerned, the incinerator ash, after removal
of ferrous and non-ferrous metals, can be ground to a fine powder, blended with clay,
pelletized and fired in a kiln to produce artificial aggregate. The material is capable of
producing concrete with compressive strengths as high as 50 MPa (7000 psi) at 28
days. There will, obviously, be problems with variations in the composition of the raw
ash, and the long-term durability characteristics of the material have yet to be
determined, although results to date look promising.
These topics are outside the scope of this book, but readers should be aware of the
new and growing possibilities of using processed waste as aggregate.

189
Chapter 4. Fresh concrete

Although fresh concrete is only of transient interest, we should note that the strength
of concrete of given mix proportions is very seriously affected by the degree of its
compaction. It is vital, therefore, that the consistency of the mix be such that the
concrete can be transported, placed, compacted, and finished sufficiently easily and
without segregation. This chapter is therefore devoted to the properties of fresh
concrete which will contribute to such an objective.
Before considering fresh concrete, we should observe that the first three chapters
discussed only two of the three essential ingredients of concrete: cement and
aggregate. The third essential ingredient is water, and this will be considered below.
It may be appropriate to add, at this stage, that many, if not most, concrete mixes
contain also admixtures: these are the topic of Chapter 5.
Quality of mixing water
The vital influence of the quantity of water in the mix on the strength of the resulting
concrete will be considered in Chapter 6. Apart from that, traditionally, those studying
concrete have shown little interest in water in the mix. Admittedly, water is necessary
to make the mix adequately workable and it is, of course, necessary, to hydrate the
cement or, as was later established, only some of the cement. Consequently, relatively
little effort was devoted to the study of the quality of water.
However, water is not just a liquid used to make concrete: it is involved in the
whole life of concrete, for good or for evil. Most actions on concrete in service, other
than loading, involve water, either pure or carrying salts or solids. The important
influences of water, in addition to those on workability and strength, are those on:
setting, hydration, bleeding, drying shrinkage, creep, ingress of salts, explosive failure
of concrete with a very low water-cement ratio, autogenous healing, staining of the
surface, chemical attack of concrete, corrosion of reinforcement, freezing and thawing,
carbonation, alkali-silica reaction, thermal properties, electrical resistivity, cavitation
and erosion, and quality of drinking water passed through concrete pipes or mortar-
lined pipes. This is a pretty exhaustive list.
As some of the influences are for good, and others for bad, one could say that
water and concrete are in a love-hate relationship; this indeed is the title of a chapter
in my book Neville on Concrete: an examination of issues in concrete
practice.4.122 Another chapter in that book is titled “Water: Cinderella Ingredient of
Concrete”.
For these reasons, the suitability of water for mixing and curing purposes should
be considered. Clear distinction must be made between the quality of mixing water
and the attack on hardened concrete by aggressive waters. Indeed, some waters which
adversely affect hardened concrete may be harmless or even beneficial when used in
mixing.4.15 The quality of curing water is considered on p. 324.
Mixing water should not contain undesirable organic substances or inorganic
constituents in excessive proportions. However, quantitative limits of harmful
constituents are not known reliably and also, because unnecessary restrictions could
be economically damaging. Some limits are specified in BS EN 1008 : 2002.
In many project specifications, the quality of water is covered by a clause saying
that water should be fit for drinking. Such water very rarely contains dissolved

190
inorganic solids in excess of 2000 parts per million (ppm), and as a rule less than
1000 ppm. For a water/cement ratio of 0.5, the latter content corresponds to a quantity
of solids representing 0.05 per cent of the mass of cement, and any effect of the
common solids would be small.
While the use of potable water as mixing water is generally satisfactory, there are
some exceptions; for instance, in some arid areas, local drinking water is saline and
may contain an excessive amount of chlorides. Also, some natural mineral waters
contain undesirable amounts of alkali carbonates and bicarbonates which could
contribute to the alkali–silica reaction.
Conversely, some waters not fit for drinking may often be used satisfactorily in
making concrete. As a rule, water with pH of 6.0 to 8.0,4.33 or possibly even 9.0, which
does not taste brackish is suitable for use, but dark colour or bad smell do not
necessarily mean that deleterious substances are present.4.16 A simple way of
determining the suitability of such water is to compare the setting time of cement and
the strength of mortar cubes using the water in question with the corresponding results
obtained using known ‘good’ water or distilled water; there is no appreciable
difference between the behaviour of distilled and ordinary drinking water. A tolerance
of about 10 per cent is usually permitted to allow for chance variations in
strength;4.15 BS EN 1008-2002 also specifies 10 per cent. Such tests are recommended
when water for which no service record is available contains dissolved solids in
excess of 2000 ppm or, in the case of alkali carbonate or bicarbonate, in excess of
1000 ppm. When unusual solids are present a test is also advisable. Limits on
chlorides, sulfates and alkali are given in BS EN 1008 : 2002 and ASTM C 1602-06.
Because it is undesirable to introduce large quantities of clay and silt into the
concrete, mixing water with a high content of suspended solids should be allowed to
stand in a settling basin before use; a turbidity limit of 2000 ppm has been
suggested.4.7 However, water used to wash out truck mixers is satisfactory as mixing
water, provided of course that it was satisfactory to begin with. ASTM C 94-94a and
BS EN 1008-2002 give the requirements for the use of wash water. Clearly, cements
and admixtures different from those originally used should not be involved. The use
of wash water is an important topic, but is outside the scope of this book.
Natural waters that are slightly acid are harmless, but water containing humic or
other organic acids may adversely affect the hardening of concrete; such water, as
well as highly alkaline water, should be tested. The effects of different ions vary, as
shown by Steinour.4.15
It may be interesting to note that the presence of algae in mixing water results in
air entrainment with a consequent loss of strength.4.13 According to the appendix to BS
3148 : 1980, green or brown slime-forming algae should be regarded with suspicion,
and water containing them should be tested.
Brackish water contains chlorides and sulfates. When chloride does not exceed 500
ppm, or SO3 does not exceed 1000 ppm, the water is harmless, but water with even
higher salt contents has been used satisfactorily.4.35 The appendix to BS 3148 : 1980
recommends limits on chloride and on SO3 as above, and also recommends that alkali
carbonates and bicarbonates should not exceed 1000 ppm. Somewhat less severe
limitations are recommended in American literature.4.33
Sea water has a total salinity of about 3.5 per cent (78 per cent of the dissolved
solids being NaCl and 15 per cent MgCl2 and MgSO4) (cf. p. 517), and produces a
slightly higher early strength but a lower long-term strength; the loss of strength is

191
usually no more than 15 per cent4.25 and can therefore often be tolerated. Some tests
suggest that sea water slightly accelerates the setting time of cement, others4.27 show a
substantial reduction in the initial setting time but not necessarily in the final set.
Generally, the effects on setting are unimportant if water is acceptable from strength
considerations. BS EN 1008-2002 specifies a tolerance of 25 minutes in the initial
setting time and a maximum final setting time of 12 hours.
Water containing large quantities of chlorides (e.g. sea water) tends to cause
persistent dampness and surface efflorescence. Such water should, therefore, not be
used where appearance of unreinforced concrete is of importance, or where a plaster
finish is to be applied.4.9 Much more importantly, the presence of chlorides in concrete
containing embedded steel can lead to its corrosion; the limits on the total chloride ion
content in concrete are considered on p. 566.
In this connection, but also with respect to all impurities in water, it is important to
remember that water discharged into the mixer is not the only source of mix water:
aggregate usually contains surface moisture (see p. 132). This water can represent a
substantial proportion of the total mixing water. It is, therefore, important that the
water brought in by the aggregate is also free from harmful material.
Tests on mixes with a range of waters suitable for use in concrete showed no effect
on the structure of the hydrated cement paste.4.103
The preceding discussion was concerned with structural concrete, usually
reinforced or prestressed. Under particular circumstances, for instance in the
construction of unreinforced concrete bulkheads in a mine, highly contaminated water
can be used. Al-Manaseer et al.4.102 showed that water containing very high percentages
of salts of sodium, potassium, calcium and magnesium used in making concrete
containing Portland cement blended with fly ash did not adversely affect the strength
of concrete. However, no information on long-term behaviour is available.
Biologically treated domestic waste water has also been investigated for use as mixing
water,4.40 but much more information about the variability of such water, health
hazards and long-term behaviour is required.
On page 183 reference was made to a possible effect of cement in the interior
surface of a pipe on water destined for human consumption. As long as water moves
through a concrete pipe (or a mortar-lined conduit) at speed, no significant chemical
reaction with cement occurs. However, when water is near-stagnant, as during the
night in domestic water conduits, leaching of cement may occur. This may raise the
pH of the water and increase the content of CaCO3, referred to as carbonate alkalinity
or water hardness. The increase in CaCO3 is induced by carbon dioxide dissolved in
the water and a reaction with calcium hydroxide by water may also increase the
content of aluminium, calcium, sodium, and potassium and of corrosion inhibitors in
the mix.4.122
Curing water should generally satisfy the requirements for mixing water, but it
should be free from substances that attack hardened concrete. Also, flowing pure
water dissolves Ca(OH)2 and causes surface erosion. Curing very young concrete with
seawater may lead to an attack on reinforcement.
Density of fresh concrete
Density, also called unit mass or unit weight in air, can be determined experimentally
by using ASTM standard C 138-09 or BS EN 12350-6 : 2009. Theoretically, density

192
is the sum of masses of all the ingredients of a batch of concrete divided by the
volume filled by the concrete.
Alternatively, knowing the density of fresh concrete, the yield per batch can be
determined as the mass of all the ingredients in a batch divided by the density.
Definition of workability
A concrete which can be readily compacted is said to be workable, but to say merely
that workability determines the ease of placement and the resistance to segregation is
too loose a description of this vital property of concrete. Furthermore, the desired
workability in any particular case would depend on the means of compaction
available; likewise, a workability suitable for mass concrete is not necessarily
sufficient for thin, inaccessible, or heavily reinforced sections. For these reasons,
workability should be defined as a physical property of concrete alone without
reference to the circumstances of a particular type of construction.
To obtain such a definition it is necessary to consider what happens when concrete
is being compacted. Whether compaction is achieved by ramming or by vibration, the
process consists essentially of the elimination of entrapped air from the concrete until
it has achieved as close a configuration as is possible for a given mix. Thus, the work
done is used to overcome the friction between the individual particles in the concrete
and also between the concrete and the surface of the mould or of the reinforcement.
These two can be called internal friction and surface friction, respectively. In addition,
some of the work done is used in vibrating the mould or in shock and, indeed, in
vibrating those parts of the concrete which have already been fully consolidated. Thus
the work done consists of a ‘wasted’ part and ‘useful’ work, the latter, as mentioned
before, comprising work done to overcome the internal friction and the surface
friction. Because only the internal friction is an intrinsic property of the mix,
workability can be best defined as the amount of useful internal work necessary to
produce full compaction. This definition was developed by Glanville et al.4.1 who
exhaustively examined the field of compaction and workability. The ASTM C 125-
09a definition of workability is somewhat more qualitative: “property determining the
effort required to manipulate a freshly mixed quantity of concrete with minimum loss
of homogeneity”. The ACI definition of workability, given in ACI 116R-90,4.46 is:
“that property of freshly mixed concrete or mortar which determines the ease and
homogeneity with which it can be mixed, placed, consolidated, and finished”.
Another term used to describe the state of fresh concrete is consistency. In ordinary
English usage, this word refers to the firmness of form of a substance or to the ease
with which it will flow. In the case of concrete, consistency is sometimes taken to
mean the degree of wetness; within limits, wet concretes are more workable than dry
concretes, but concretes of the same consistency may vary in workability. The ACI
definition of consistency is: “the relative mobility or ability of freshly mixed concrete
or mortar to flow”;4.46 this is measured by slump.
Technical literature abounds with variations of the definitions of workability and
consistency but they are all qualitative in nature and more reflections of a personal
viewpoint rather than of scientific precision. The same applies to the plethora of terms
such as: flowability, mobility, and pumpability. There is also a term ‘stability’ which
refers to the cohesion of the mix, that is, its resistance to segregation. These terms do
have specific meaning but only under a set of given circumstances; they can rarely be
used as an objective and quantifiable description of a concrete mix.

193
 A good review of the attempts to define the various terms is presented by
Bartos,4.56 among others.
The need for sufficient workability
Workability has so far been discussed merely as a property of fresh concrete: it is,
however, also a vital property as far as the finished product is concerned because
concrete must have a workability such that compaction to maximum density is
possible with a reasonable amount of work or with the amount that we are prepared to
put in under given conditions.
The need for compaction becomes apparent from a study of the relation between
the degree of compaction and the resulting strength. It is convenient to express the
former as a density ratio, i.e. a ratio of the actual density of the given concrete to the
density of the same mix when fully compacted. Likewise, the ratio of the strength of
the concrete is actually (partially) compacted to the strength of the same mix when
fully compacted can be called the strength ratio. Then the relation between the
strength ratio and the density ratio is of the form shown in Fig. 4.1. The presence of
voids in concrete greatly reduces its strength: 5 per cent of voids can lower strength
by as much as 30 per cent, and even 2 per cent voids can result in a drop of strength of
more than 10 per cent.4.1 This, of course, is in aggreement with Féret’s expression
relating strength to the sum of the volumes of water and air in the hardened cement
paste (see p. 271).

Fig. 4.1. Relation between strength ratio and density ratio.4.1 (Crown copyright)
Voids in concrete are in fact either bubbles of entrapped air or spaces left after
excess water has been removed. The volume of the latter depends primarily on the
water/cement ratio of the mix; to a lesser extent, there may be spaces arising from
water trapped underneath large particles of aggregate or underneath reinforcement.
The air bubbles, which represent ‘accidental’ air, i.e. voids within an originally loose
granular material, are governed by the grading of the finest particles in the mix and
are more easily expelled from a wetter mix than from a dry one. It follows, therefore,
that for any given method of compaction there may be an optimum water content of
the mix at which the sum of the volumes of air bubbles and water space will be a
minimum. At this optimum water content, the highest density ratio of the concrete
would be obtained. It can be seen, however, that the optimum water content may vary
for different methods of compaction.
Factors affecting workability

194
The main factor is the water content of the mix, expressed in kilograms (or litres) of
water per cubic metre of concrete: it is convenient, though approximate, to assume
that, for a given type and grading of aggregate and workability of concrete, the water
content is independent of the aggregate/cement ratio or of the cement content of the
mix. On the basis of this assumption, the mix proportions of concretes of different
richness can be estimated, and Table 4.1 gives typical values of water content for
different slumps and maximum sizes of aggregate. These values are applicable to non-
air-entrained concrete only. When air is entrained, the water content can be reduced in
accordance with the data of Fig. 4.2.4.2 This is indicative only, because the effect of
entrained air on workability depends on the mix proportions, as described in detail on
p. 562.

Table 4.1. Approximate Water Content for Different Slumps and Maximum
Sizes of Aggregate (partially based on the approach of the National Aggregates
Association in the United States)

195
 Fig. 4.2. Reduction in mixing water requirement due to addition of air by air
entrainment4.2
If the water content and the other mix proportions are fixed, workability is
governed by the maximum size of aggregate, its grading, shape and texture. The
influence of these factors was discussed in Chapter 3. However, the grading and the
water/cement ratio have to be considered together, as a grading producing the most
workable concrete for one particular value of water/cement ratio may not be the best
for another value of the ratio. In particular, the higher the water/cement ratio the finer
the grading required for the highest workability. In actual fact, for a given value of
water/cement ratio, there is one value of the coarse/fine aggregate ratio (using given
materials) that gives the highest workability.4.1 Conversely, for a given workability,
there is one value of the coarse/fine aggregate ratio which needs the lowest water
content. The influence of these factors was discussed in Chapter 3.
It should be remembered, however, that, although, when discussing gradings of
aggregate required for a satisfactory workability, proportions by mass were laid down,
these apply only to aggregate of a constant specific gravity. In actual fact, workability
is governed by the volumetric proportions of particles of different sizes, so that when
aggregates of varying specific gravity are used (e.g. in the case of some lightweight
aggregates or mixtures of ordinary and lightweight aggregates) the mix proportions
should be assessed on the basis of absolute volume of each size fraction. This applies
also in the case of air-entrained concrete because the entrained air behaves like
weightless fine particles. An example of a calculation on absolute volume basis is
given on p. 747. The influence of the properties of aggregate on workability decreases

196
with an increase in the richness of the mix, and possibly disappears altogether when
the aggregate/cement ratio is as low as or 2.
In practice, predicting the influence of mix proportions on workability requires
care since, of the three factors, water/cement ratio, aggregate/cement ratio and water
content, only two are independent. For instance, if the aggregate/cement ratio is
reduced, but the water/cement ratio is kept constant, the water content increases, and
consequently the workability also increases. If, on the other hand, the water content is
kept constant when the aggregate/cement ratio is reduced, then the water/cement ratio
decreases but workability is not seriously affected.
The last qualification is necessary because of some secondary effects: a lower
aggregate/cement ratio means a higher total surface area of solids (aggregate and
cement) so that the same amount of water results in a somewhat
decreased workability. This could be offset by the use of a slightly coarser grading of
aggregate. There are also other minor factors such as fineness of cement, but the
influence of this is still controversial.
Measurement of workability
Unfortunately, there is no acceptable test which will measure directly the workability
as given by any of the definitions on p. 187. Numerous attempts have been made,
however, to correlate workability with some easily determinable physical
measurement, but none of these is fully satisfactory although they may provide useful
information within a range of variation in workability.
Slump test
This is a test used extensively in site work all over the world. The slump test does not
measure the workability of concrete, although ACI 116R-904.46 describes it as a
measure of consistency, but the test is very useful in detecting variations in the
uniformity of a mix of given nominal proportions.
The slump test is prescribed by ASTM C 143-10 and BS 1881 : 103 : 1993. The
mould for the slump test is a frustum of a cone, 300 mm (12 in.) high. It is placed on a
smooth surface with the smaller opening at the top, and filled with concrete in three
layers. Each layer is tamped 25 times with a standard 16 mm ( in.) diameter steel rod,
rounded at the end, and the top surface is struck off by means of a sawing and rolling
motion of the tamping rod. The mould must be firmly held against its base during the
entire operation; this is facilitated by handles or foot-rests brazed to the mould.
Immediately after filling, the cone is slowly lifted, and the unsupported concrete
will now slump – hence the name of the test. The decrease in the height of the
slumped concrete is called slump, and is measured to the nearest 5 mm ( in.). The
decrease is measured to the highest point according to BS EN 12350-2 : 2009, but to
the “displaced original center” according to ASTM C 143-10. In order to reduce the
influence on slump of the variation in the surface friction, the inside of the mould and
its base should be moistened at the beginning of every test, and prior to lifting of the
mould the area immediately around the base of the cone should be cleaned of concrete
which may have dropped accidentally.
If instead of slumping evenly all round as in a true slump (Fig. 4.3), one half of the
cone slides down an inclined plane, a shear slump is said to have taken place, and the

197
test should be repeated. If shear slump persists, as may be the case with harsh mixes,
this is an indication of lack of cohesion in the mix.

Fig. 4.3. Slump: true, shear, and collapse

Mixes of stiff consistency have a zero slump, so that, in the rather dry range, no
variation can be detected between mixes of different workability. Rich mixes behave
satisfactorily, their slump being sensitive to variations in workability. However, in a
lean mix with a tendency to harshness, a true slump can easily change to the shear
type, or even to collapse (Fig. 4.3), and widely different values of slump can be
obtained in different samples from the same mix.
The approximate magnitude of slump for different workabilities (in a modified
form of Bartos’ proposals4.56) is given in Table 4.2. Table 4.3 gives the proposed
European classification of BS EN 206-1 : 2000. One reason for the
difference between the two tables is that the European approach is to measure slump
to the nearest 10 mm. It should be remembered, however, that with different
aggregates, especially a different content of fine aggregate, the same slump can be
recorded for different workabilities, as indeed the slump bears no unique relation to
the workability as defined earlier. Moreover, slump does not measure the ease of
compaction of concrete and, as slump occurs under the self-weight of the test concrete
only, it does not reflect behaviour under dynamic conditions such as vibration,
finishing, pumping or moving through a tremie. Rather, slump reflects the ‘yield’ of
concrete.4.110

Table 4.2. Description of Workability and Magnitude of Slump

Table 4.3. Classification of Workability and Magnitude of Slump According to

BS EN 206-1 : 2000

198
 Despite these limitations, the slump test is very useful on the site as a check on the
batch-to-batch or hour-to-hour variation in the materials being fed into the mixer. An
increase in slump may mean, for instance, that the moisture content of aggregate has
unexpectedly increased; another cause would be a change in the grading of the
aggregate, such as a deficiency of sand. Too high or too low a slump gives immediate
warning and enables the mixer operator to remedy the situation. This application of
the slump test, as well as its simplicity, is responsible for its widespread use.
A mini-slump test was developed for the purpose of assessing the influence of
various water-reducing admixtures and superplasticizers on neat cement paste.4.105 The
test may be useful for that specific purpose, but it is important to remember that the
workability of concrete is affected also by factors other than the flow properties of the
constituent cement paste.
Compacting factor test
There is no generally accepted method of directly measuring the amount of work
necessary to achieve full compaction, which is a definition of workability.4.1 Probably
the best test yet available uses the inverse approach: the degree of compaction
achieved by a standard amount of work is determined. The work applied includes
perforce the work done against the surface friction but this is reduced to a minimum,
although probably the actual friction varies with the workability of the mix.
The degree of compaction, called the compacting factor, is measured by the
density ratio, i.e. the ratio of the density actually achieved in the test to the density of
the same concrete fully compacted.
The test, known as the compacting factor test, is described in BS 1881-103 : 1993
and in ACI 211.3-75 (Revised 1987) (Reapproved 1992),4.70 and is appropriate for
concrete with a maximum size of aggregate up to 40 mm (or in.). The apparatus
consists essentially of two hoppers, each in the shape of a frustum of a cone, and one
cylinder, the three being above one another. The hoppers have hinged doors at the
bottom, as shown in Fig. 4.4. All inside surfaces are polished to reduce friction.

199
 Fig. 4.4. Compacting factor apparatus
The upper hopper is filled with concrete, this being placed gently so that at this
stage no work is done on the concrete to produce compaction. The bottom door of the
hopper is then released and the concrete falls into the lower hopper. This is smaller
than the upper one and is, therefore, filled to overflowing, and thus always contains
approximately the same amount of concrete in a standard state; this reduces the
influence of the personal factor in filling the top hopper. The bottom door of the lower
hopper is then released and the concrete falls into the cylinder. Excess concrete is cut
by two floats slid across the top of the mould, and the net mass of concrete in the
known volume of the cylinder is determined.
The density of the concrete in the cylinder is now calculated, and this density
divided by the density of the fully compacted concrete is defined as the compacting
factor. The latter density can be obtained by actually filling the cylinder with concrete
in four layers, each tamped or vibrated, or alternatively calculated from the absolute
volumes of the mix ingredients. The compacting factor can also be calculated from
the reduction in volume that occurs when a defined volume of partially compacted
concrete (by passing through the hoppers) is fully compacted.
The compacting factor apparatus shown in Fig. 4.4 is about 1.2 m (4 ft) high and
its use is generally limited to pavement construction and precast concrete manufacture.

200
 Table 4.4 lists values of the compacting factor for different workabilities.4.3 Unlike
the slump test, variations in the workability of dry concrete are reflected in a large
change in the compacting factor, i.e. the test is more sensitive at the low workability
end of the scale than at high workability. However, very dry mixes tend to stick in one
or both hoppers and the material has to be eased gently by poking with a steel rod.
Moreover, it seems that for concrete of very low workability the actual amount of
work required for full compaction depends on the richness of the mix while the
compacting factor does not: leaner mixes need more work than richer ones.4.4 This
means that the implied assumption that all mixes with the same compacting factor
require the same amount of useful work is not always justified. Likewise, the
assumption, mentioned earlier, that the wasted work represents a constant proportion
of the total work done regardless of the properties of the mix is not quite correct.
Nevertheless, the compacting factor test undoubtedly provides a good measure of
workability.

Table 4.4. Description of Workability and Compacting Factor4.3

ASTM flow test

This laboratory test gives an indication of the consistency of concrete and its
proneness to segregation by measuring the spread of a pile of concrete on a
table subjected to jolting. This test also gives a good assessment of consistency of stiff,
rich, and rather cohesive mixes. The test was covered by ASTM C 124-39
(Reapproved 1966) which was withdrawn in 1974 because the test was little used,
rather than because it was thought to be not appropriate.
Remoulding test
Use of a jolted table is made in another test, in which an assessment of workability is
made on the basis of the effort involved in changing the shape of a sample of concrete.
This is the remoulding test, developed by Powers.4.5
The apparatus is shown diagrammatically in Fig. 4.5. A standard slump cone is
placed in a cylinder 305 mm (12 in.) in diameter and 203 mm (8 in.) high, the cylinder
being mounted rigidly on a flow table, adjusted to give a 6.3 mm ( in.) drop. Inside
the main cylinder, there is an inner ring, 210 mm ( in.) in diameter and 127 mm (5
in.) high. The distance between the bottom of the inner ring and the bottom of the
main cylinder can be set between 67 and 76 mm ( and 3 in.).

201
 Fig. 4.5. Remoulding test apparatus
The slump cone is filled in the standard manner, removed, and a disc-shaped rider
(weighing 1.9 kg (4.3 lb)) is placed on top of the concrete. The table is now jolted at
the rate of one jolt per second until the bottom of the rider is 81 mm ( in.) above
the base plate. At this stage, the shape of the concrete has changed from a frustum of a
cone to a cylinder. The effort required to achieve this remoulding is expressed as the
number of jolts required. For very dry mixes a considerable effort may be necessary.
The test is purely a laboratory one but is valuable because the remoulding effort
appears to be closely related to workability.
Vebe test
This is a development of the remoulding test in which the inner ring of Powers’
apparatus is omitted and compaction is achieved by vibration instead of jolting. The
apparatus is shown diagrammatically in Fig. 4.6. The name ‘Vebe’ is derived from the
initials of V. Bährner of Sweden who developed the test. The test is covered by BS
EN 12350-3 : 2009; it is referred to also in ACI 211.3-75 (Revised 1987).4.70

202
 Fig. 4.6. Vebe apparatus
The remoulding is assumed to be complete when the glass plate rider is completely
covered with concrete and all cavities in the surface of the concrete have disappeared.
This is judged visually, and the difficulty of establishing the end point of the test may
be a source of error. To overcome it, an automatically operated device for recording
the movement of the plate against time may be fitted.
Compaction is achieved using a vibrating table with an eccentric mass rotating at
50 to 60 Hz and a maximum acceleration of 3g to 4g. It is assumed that the input of
energy required for compaction is a measure of workability of the mix, and this is
expressed as the time in seconds, called Vebe time, required for the remoulding to be
complete. Sometimes, a correction for the change in the volume of concrete
from V2 before, to V1 after, vibration is applied, the time being multiplied by V2/V1. The
test is appropriate for mixes with a Vebe time between 3 and 30 seconds.
Vebe is a good laboratory test, particularly from very dry mixes. This is in contrast
to the compacting factor test where error may be introduced by the tendency of some
dry mixes to stick in the hoppers. The Vebe test also has the additional advantage that
the treatment of concrete during the test is comparatively closely related to the
method of placing in practice. Both the Vebe test and the remoulding tests determine
the time required to achieve compaction, which is related to the total work done.

203
Flow table test
This test, which was developed in Germany in 1933, was covered by BS 1881 : 105 :
1984. The test is appropriate for concrete of high and very high workability, including
flowing concrete (see p. 259) which would exhibit a collapse slump.
The apparatus consists essentially of a wooden board covered by a steel plate with
a total mass of 16 kg. This board is hinged along one side to a base board, each board
being 700 mm square. The upper board can be lifted up to a stop so that the free edge
rises 40 mm. Appropriate markings indicate the location of the concrete to be
deposited on the table.
The table top is moistened and a frustum of a cone of concrete, lightly tamped by a
wooden tamper in a prescribed manner, is placed using a mould 200 mm high with a
bottom diameter of 200 mm and a top diameter of 130 mm. Excess concrete is
removed, the surrounding table top is cleaned and, after an interval of 30 seconds, the
table top is lifted 15 times in a period of 45 to 75 seconds, this motion avoiding a
significant force against the stop. In consequence, the concrete spreads and the
maximum spread parallel to the two edges of the table is measured. The average of
these two values, given to the nearest millimetre, represents the flow. The test is
appropriate for mixes having a flow of 340 to 600 mm. If the concrete at this stage
does not appear uniform and cohesive, this is an indication of a lack of cohesiveness
of the mix. The current standard is BS EN 12350-5 : 2010.
A laboratory investigation4.39 has shown a linear relation between flow and slump,
but the tests were limited in scope in that they involved only one aggregate type and
only one aggregate grading. Also, the effect of site conditions was not included. In
consequence, no generalization can be inferred from the data which were published,
and it would be unwise to view the slump test and the flow test as generally
interchangeable. In essence, the two tests do not measure the same physical
phenomena so that there is no reason to expect a single relationship between the two
when grading or aggregate shape or content of fine material in the mix vary. For
practical purposes, an appropriate test should be adopted. Such a test makes it
possible to recognize a departure from the specified mix proportions, and this is what
matters on site.
Ball penetration test and compactability test
This is a simple field test consisting of the determination of the depth to which a 152
mm (6 in.) diameter metal hemisphere, weighing 13.6 kg (30 lb), will sink under its
own weight into fresh concrete. A sketch of the apparatus, devised by J. W. Kelly and
known as the Kelly ball, is shown in Fig. 4.7.

204
 Fig. 4.7. Kelly ball
The use of this test is similar to that of the slump test, that is, routine checking of
consistency for control purposes. The test is essentially an American one, and is rarely
used elsewhere. It is, however, worth considering the Kelly ball test as an alternative
to the slump test, over which it has some advantages. In particular, the ball test is
simpler and quicker to perform and, what is more important, it can be applied to
concrete in a buggy or actually in the form. In order to avoid the effects of a boundary,
the depth of the concrete being tested should be not less than 200 mm (8 in.), and the
least lateral dimension should be 460 mm (18 in.).
As would be expected, there is no simple correlation between penetration and
slump, since neither test measured any basic property of concrete but only the
response to specific conditions. On a site, when a particular mix is used, correlation
can be found, as shown for instance in Fig. 4.8.4.6 In practice, the ball test is essentially
used to measure variations in the mix, such as those due to a variation in the moisture
content of the aggregate.

205
 Fig. 4.8. Relation between Kelly ball penetration and slump4.6
A compactability test was introduced by BS EN 12350-4 : 2009, which determines
the reduction in volume of a loosely-packed concrete after vibration in a cylinder.
The degree of compactability is the ratio of the height of the cylinder to the height of
the compacted concrete. The compaction is effected by a vibrating table or by an
internal vibrator.
Nasser’s K-tester
Among the various attempts to devise a simple workability test, the probe test of
Nasser4.41 deserves mention. This test uses a hollow probe 19 mm ( in.) in diameter
with openings through which mortar can enter the tube. The probe is inserted
vertically into fresh concrete in situ (and thus avoids using a sample). The height of
the mortar in the tube after 1 minute and also the residual height following withdrawal
of the probe are measured.
It is claimed,4.42,4.106 that these readings give an indication of consistency and
workability of the concrete because the probe readings are affected by cohesive,
adhesive, and friction forces within the mix. Thus, an over-wet mix, which exhibits a
high slump, would lead to a relatively low level of mortar retained in the probe, this
being the result of segregation. The residual height of mortar in the probe appears to
be related to slump, providing this does not exceed 80 mm (or 3 in.).4.41 However, the
K-tester can be used even for flowing concrete.4.106 The K-tester has not been
standardized and is not widely used.
Two-point test

206
Tattersall4.43 has repeatedly criticized all the existing workability tests on the grounds
that they measure only one parameter. His argument is that the flow of fresh concrete
should be described by the Bingham model, i.e. by the equation

where τ = shear stress at rate of shear

τ0 = yield stress and
μ = plastic viscosity.
Because there are two unknowns, measurements at two rates of shear are required;
hence, the name ‘two-point test’. The yield stress represents the threshold value for
flow to begin and is closely related to slump.4.107 The plastic viscosity reflects the
increase in shear stress with an increase in the rate of shear.
Tattersall4.43 developed techniques of torque measurement using a modified food
mixer. Hence, he deduced experimentally data related to the shear stress at a given
rate of shear and to constants representing the yield stress, τ0, and plastic viscosity, μ,
of the mix. It is the latter two that, in his view, provide a measure of the fundamental
rheological properties of concrete. Their determination requires the measurement of
torque to rotate the mixer at two speeds. This apparatus was modified both by
Tattersall4.43 and by Wallevik and Gjørv4.104 who claim that their apparatus is more
reliable and, in addition, gives a quantified measure of the susceptibility of the mix to
segregation.
Problems in use are that the apparatus is cumbersome, complicated, and requires
skill in interpretation of the test readings, which are not directly usable, unlike slump.
For these reasons, the two-point test is inappropriate for site operation as a means of
control, but may be of value in the laboratory.
With respect to two-point description of workability, it is worth noting that, for
robot-placed concrete, it is important to establish the value of plastic viscosity and
yield stress of concrete, and the variation in these two parameters with temperature
and time since mixing. Equations predicting viscosity on the basis of the viscosity
equation for high-concentration suspensions, taking into account aggregate properties
and using experimental constants, were developed by Murata and Kikukawa.4.107 They
also developed an equation for the yield value of concrete based on slump. The
validity of this approach is yet to be proven.
Comparison of tests
It should be said at the outset that no comparison is really possible as each test
measures the behaviour of concrete under different conditions. The particular uses of
each test have been mentioned but it is worth adding that BS 1881 : 1983 (withdrawn
but useful) lists the test methods appropriate to mixes of different workability as
shown in Table 4.5.

Table 4.5. Test Methods Apropriate to Mixes of Different Workability According

to BS 1881 : 1983

207
 The compacting factor test is closely related to the reciprocal of workability,
whereas the remoulding, flow, and Vebe tests are direct functions of workability. The
Vebe test measures the properties of concrete under vibration as compared with the
free-fall conditions of the compacting factor test and the jolting in the remoulding and
flow tests. All four tests are satisfactory in the laboratory, but the compacting factor
apparatus is also suitable for site use.
An indication of the relation between the compacting factor and the Vebe time is
given by Fig. 4.9, but this applies only to the mixes used, and the relation must not be
assumed to be generally applicable because it depends on factors such as the shape
and texture of the aggregate or presence of entrained air, as well as on mix
proportions. For specific mixes, the relation between compacting factor and slump has
been obtained, but such a relation is also a function of the properties of the mix. The
relation between the number of jolts in Powers’ remoulding test and slump (Fig. 4.10)
is also only broadly defined.4.58 A general indication of the pattern of the relation
between the compacting factor, Vebe time and slump is shown in Fig. 4.11.4.14 The
influence of the richness of the mix in two of these relations is clear. The absence of
influence in the case of the relation between slump and Vebe time is illusory because
slump is insensitive at one end of the scale (low workability) and Vebe time is
insensitive at the other end; thus two asymptotic lines with a small connecting part are
present.

208
 Fig. 4.9. Relation between compacting factor and Vebe time4.4

209
 Fig. 4.10. Relation between the number of jolts using Powers’ remoulding test
apparatus and slump for mixes with fine aggregates of different fineness4.58

210
 Fig. 4.11. General pattern of relations between workability tests for mixes of
varying aggregate/cement ratios4.14
The flow test is valuable in assessing the cohesiveness and workability of very
high workability concrete or flowing concrete.
The slump and penetration tests are purely comparative and, in that capacity, both
are very useful except that the slump test is unreliable with lean mixes, for which

211
good control is often of considerable importance. The slump test is periodically
attacked as useless and as a poor indicator of the strength of concrete.4.52,4.111 Such
criticism may well be misplaced because the slump test does not purport to measure
the potential strength of concrete: the purpose of the slump test is to verify the
uniformity of the slump from batch to batch; and no more. Such a verification is
useful in that it ensures that the concrete, as placed, has the desired workability.
Moreover, the mere knowledge that testing is under way concentrates the mind at the
batching plant, and the psychological effect of this knowledge is to prevent a lapse
into the ‘anything-will-do’ attitude.
It has to be admitted that the slump test, which represents a single rate of shear
situation, cannot fully characterize the workability of concrete. The test can, however,
give a comparative value of workability if the only variable is the water content of the
mix because, under such circumstances, the straight lines representing the Bingham
equations do not cross one another.4.43 A perfect, practical test for workability has yet
to be devised. Although this seems primitive, there is value in visual assessment of
workability by patting concrete with a trowel in order to see the ease of finishing.
Experience is clearly necessary but, once it has been acquired, the ‘by eye’ test,
particularly for the purpose of checking uniformity, is both rapid and reliable.
Stiffening time of concrete
It is possible to determine whether concrete has stiffened to a given degree by testing
mortar sieved out of the concrete, using a 5 mm (No. 4 ASTM) sieve. A spring
reaction-type probe, known as Proctor probe, is used to determine the times when the
resistance to penetration is 3.5 MPa (500 psi) and 27.6 MPa (4000 psi). The former is
referred to as initial setting time and indicates that the concrete has become too stiff to
be made mobile by vibration. The time when the resistance to penetration has reached
27.6 MPa (4000 psi) is the final setting time; the compressive strength of concrete
measured on a standard cylinder is then about 0.7 MPa (100 psi). These setting times
are distinct from the setting times of cement.
The test method is prescribed by ASTM C 403-08 and can be used for comparative
purposes. It cannot be an absolute measure because the test is performed on mortar
and not on the parent concrete. British Standard BS 5075-1 : 1982 (superseded by BS
EN 480 and 934) also prescribes a stiffening time test.
Effect of time and temperature on workability
Freshly mixed concrete stiffens with time. This should not be confused with setting of
cement. It is simply that some water from the mix is absorbed by the aggregate if not
saturated, some is lost by evaporation, particularly if the concrete is exposed to sun or
wind, and some is removed by the initial chemical reactions. The compacting factor
decreases by up to about 0.1 during a period of one hour from mixing.
The exact value of the loss in workability depends on several factors. First, the
higher the initial workability the greater the slump loss. Second, the rate of loss of
slump is higher in rich mixes. Furthermore, the rate of loss depends on the properties
of the cement used: the rate is higher when the alkali content is high4.108 and when the
sulfate content is too low.4.62 An example of the slump-time relation for concrete made
with a water/cement ratio of 0.4 and cement having an alkali content of 0.58 is shown
in Fig. 4.12.4.60

212
 Fig. 4.12. Loss of slump with time since mixing (based on ref. 4.60)
The change in workability with time depends also on the moisture condition of
aggregate (at a given total water content): the loss is greater with dry aggregate due to
the absorption of water by aggregate, as of course would be expected. Water-reducing
admixtures, although they delay the initial stiffening of concrete, often lead to a
somewhat increased rate of loss of slump with time.
The workability of a mix is also affected by the ambient temperature, although,
strictly speaking, we are concerned with the temperature of the concrete itself. Figure
4.13 gives an example of the effect of temperature on slump of laboratory-mixed
concrete:4.7 it is apparent that on a hot day the water content of the mix would have to
be increased for a constant early workability to be maintained. The loss of slump in
stiff mixes is less influenced by temperature because such mixes are less affected by
changes in water content. Figure 4.14 shows that as the concrete temperature
increases the percentage increase in water required to effect a 25 mm (1 in.) change in
slump also increases.4.8 The loss of slump with time is also affected by the temperature,
as shown in Fig. 4.15.

213
 Fig. 4.13. Influence of temperature on slump of concretes with different
maximum aggregate size4.7

Fig. 4.14. Influence of temperature on the amount of water required to change

slump4.8

214
Fig. 4.15. Influence of temperature on loss of slump after 90 minutes for concrete
with a cement content of 306 kg/m3 (517 lb/yd3) (based on ref. 4.61)
The effects of temperature on concrete are discussed in Chapter 8.
Because workability decreases with time, it is important to measure, say, slump
after a predetermined time lapse since mixing. There is value in determining slump
immediately after the discharge of the concrete from the mixer for the purpose of
control of batching. There is also value in determining slump at the time of placing
the concrete in the formwork for the purpose of ensuring that the workability is
appropriate for the means of compaction to be used.
Segregation

215
In discussing workable concrete in general terms, it was stated that such concrete
should not easily segregate, i.e. it ought to be cohesive. However, strictly speaking,
the absence of a tendency to segregate is not included in the definition of a workable
mix. Nevertheless, the absence of appreciable segregation is essential as full
compaction of a segregated mix is impossible.
Segregation can be defined as separation of the constituents of a heterogeneous
mixture so that their distribution is no longer uniform. In the case of concrete, it is the
differences in the size of particles and in the specific gravity of the mix constituents
that are the primary causes of segregation, but its extent can be controlled by the
choice of suitable grading and by care in handling.
It is worth noting that a higher viscosity of the fresh cement paste component
militates against the downward movement of the heavier aggregate particles;
consequently, mixes with low water/cement ratios are less prone to segregation.4.48
There are two forms of segregation. In the first, the coarser particles tend to
separate out because they tend to travel further along a slope or to settle more than
finer particles. The second form of segregation, occurring particularly in wet mixes, is
manifested by the separation of grout (cement plus water) from the mix. With some
gradings, when a lean mix is used, the first type of segregation may occur if the mix is
too dry; addition of water would improve the cohesion of the mix, but when the mix
becomes too wet the second type of segregation would take place.
The influence of grading on segregation was discussed in detail in Chapter 3, but
the actual extent of segregation depends on the method of handling and placing of
concrete. If the concrete does not have far to travel and is transferred directly from the
skip or bucket to the final position in the form, the danger of segregation is small. On
the other hand, dropping concrete from a considerable height, passing along a chute,
particularly with changes of direction, and discharging against an obstacle – all these
encourage segregation so that under such circumstances a particularly cohesive mix
should be used. With a correct method of handling, transporting and placing, the
likelihood of segregation can be greatly reduced: there are many practical rules, which
are presented in ACI 304R-85.4.79
It should be stressed, however, that concrete should always be placed direct in the
position in which it is to remain and must not be allowed to flow or be worked along
the form. This prohibition includes the use of a vibrator to spread a heap of concrete
over a larger area. Vibration provides a most valuable means of compacting concrete
but, because a large amount of work is being done on the concrete, the danger of
segregation (in placing as distinct from handling) due to an improper use of a vibrator
is increased. This is particularly so when vibration is allowed to continue too long:
with many mixes, separation of coarse aggregate toward the bottom of the form and
of the cement paste towards the top may result. Such concrete would obviously be
weak, and the laitance (scum) on its surface would be too rich and too wet so that a
crazed surface with a tendency to dusting might result. Laitance should be
distinguished from bleed water, which is considered in the next section.
It may be noted that entrained air reduces the danger of segregation. On the other
hand, the use of coarse aggregate whose specific gravity differs appreciably from that
of the fine aggregate would lead to increased segregation.
Segregation is difficult to measure quantitatively, but is easily detected when
concrete is handled on a site in any of the ways listed earlier as undesirable. A good
picture of cohesion of the mix is obtained by the flow test. The jolting applied during

216
the test encourages segregation, and if the mix is not cohesive the larger particles of
aggregate will separate out and move toward the edge of the table. Another form of
segregation is possible: in a sloppy mix the cement paste tends to run away from the
centre of the table leaving the coarser material behind.
As far as proneness to segregation on over-vibration is concerned, a good test is to
vibrate a concrete cylinder or cube for about 10 minutes and then to strip it and
observe the distribution of coarse aggregate: any segregation will be easily seen.
Bleeding
Bleeding, known also as water gain, is a form of segregation in which some of the
water in the mix tends to rise to the surface of freshly placed concrete. This is caused
by the inability of the solid constituents of the mix to hold all of the mixing water
when they settle downwards, water having the lowest specific gravity of all the mix
constituents. We are thus dealing with subsidence, and Powers4.10 treats bleeding as a
special case of sedimentation. Bleeding can be expressed quantitatively as the total
settlement per unit height of concrete or as a percentage of the mixing water; in
extreme cases, this may reach 20 per cent.4.112 ASTM C 232-09 prescribes two methods
of determination of total bleeding. The rate of bleeding can also be determined
experimentally.
The initial bleeding proceeds at a constant rate, but subsequently the rate of
bleeding decreases steadily. Bleeding of concrete continues until the cement paste has
stiffened sufficiently to put an end to the process of sedimentation.
If the bleeding water is remixed during finishing of the top surface, a weak
wearing surface, consisting of laitance, will be formed. This can be avoided by
delaying the finishing operations until the bleed water has evaporated, and also by the
use of wood floats and avoidance of overworking the surface. On the other hand, if
evaporation of water from the surface of the concrete is faster than the bleeding rate,
plastic shrinkage cracking may result (see p. 424).
Some of the rising water becomes trapped on the underside of coarse aggregate
particles or of reinforcement, thus creating zones of poor bond. This water leaves
behind air pockets or lenses, and because all the voids are oriented in the same
direction, the permeability of the concrete in a horizontal plane may be increased.
Hence, ingress of an attacking medium into concrete is facilitated. A horizontal zone
of weakness may also be created. The formation of such zones was confirmed by
means of tensile tests in the direction of casting and at right angles to it.4.65 Trapping an
appreciable amount of bleed water must be avoided also because of the danger of frost
damage, especially in road slabs.
Some bleeding is unavoidable. However, in high elements, such as columns or
walls, as bleed water moves upwards, the water/cement ratio in the lower part of the
element is reduced, but the water trapped in the upper part of the now stiffer concrete
results in an increased water/cement ratio there, and hence in a reduced strength (see
p. 272).
The bleed water can also travel upwards along the surface of the form; if a channel
is formed due to some imperfection in the form surface, a preferred drainage path is
created with resulting surface streaking. Vertical bleed channels can also form in the
interior of the concrete.
Bleeding need not necessarily be harmful. If the surface of the concrete is to be
vacuum-dewatered (see p. 234) the removal of water is facilitated. If bleeding is

217
undisturbed and the water evaporates, the effective water/cement ratio may be
lowered with a resulting increase in strength. On the other hand, if the rising water
carries with it a considerable amount of the finer cement particles, a layer of laitance
will be formed. If this is at the top of a slab, a porous and weak surface layer will
result, with a permanently ‘dusty’ surface. At the top of a lift, a plane of weakness
would form and the bond with the next lift would be inadequate. For this reason,
laitance should always be removed by brushing and washing.
The tendency to bleeding depends largely on the properties of cement. Bleeding is
decreased by increasing the fineness of cement, possibly because finer particles
hydrate earlier and also because their rate of sedimentation is lower. Other properties
of cement also affect bleeding: there is less bleeding when the cement has a high
alkali content, a high C3A content, or when calcium chloride is added;4.11 for
limitations on the use of calcium chloride see p. 566. The test methods for bleeding of
cement pastes and mortar were prescribed by ASTM C 243-85 (withdrawn).
The properties of cement, however, are not the sole factor influencing the bleeding
of concrete4.120 so that other factors must also be considered. Specifically, the presence
of an adequate proportion of very fine aggregate particles (especially smaller than
150 μm (No. 100 sieve)) significantly reduces bleeding.4.12 The use of crushed fine
aggregate does not necessarily lead to more bleeding than rounded sand. In fact, when
the crushed fine aggregate contains excess very fine material (up to about 15 per cent
passing the 150 μm (No. 100) sieve), bleeding is reduced,4.37 but the very fine material
must consist of crusher dust only, and not of clay.
Rich mixes are less prone to bleeding than lean ones. Reduction in bleeding is
obtained by the addition of pozzolanas or other fine material or aluminium powder.
Schiessl and Schmidt4.66 found that addition to mortar of fly ash or silica fume
significantly decreased bleeding. This may not necessarily be so in the case of
concrete, much depending on the basis of comparison, e.g. whether the cementitious
materials are additional to Portland cement or whether they replace some of it. Air
entrainment effectively reduces bleeding so that finishing can follow casting without
delay.
A higher temperature, within the normal range, increases the rate of bleeding, but
the total bleeding capacity is probably unaffected. Very low temperature, however,
may increase the bleeding capacity, probably because there is more time prior to
stiffening for bleeding to occur.4.68
The influence of admixtures is not straightforward. Superplasticizers generally
decrease bleeding except at a very high slump.4.67 However, if they are used with a
retarder, increased bleeding may occur,4.68 possibly because retardation allows more
time for bleeding to occur. If, at the same time, air entrainment is used, its effect in
reducing bleeding may be dominant.
The mixing of concrete
It is essential that the mix ingredients, whose properties were discussed in Chapters
1 to 3, are properly mixed so as to produce fresh concrete in which the surface of all
aggregate particles is coated with cement paste and which is homogeneous on the
macro-scale and therefore possessing uniform properties. Almost invariably, mixing
is effected by mechanical mixers.
Concrete mixers

218
Concrete mixers must not only achieve the uniformity of the mix, just referred to, but
they must also discharge the mix without disturbing that uniformity. In fact, the
method of discharging is one of the bases of classification of concrete mixers. Several
types exist. In the tilting mixer, the mixing chamber, known as the drum, is tilted for
discharging. In the non-tilting mixer, the axis of the mixer is always horizontal, and
discharge is obtained either by inserting a chute into the drum or by reversing the
direction of rotation of the drum (when the mixer is known as a reversing drum
mixer), or rarely by splitting of the drum. There are also pan-type mixers, rather
similar in operation to an electric cake-mixer; these are called forced action mixers, as
distinct from the tilting and non-tilting mixers which rely on the free fall of concrete
in the drum.
Tilting mixers usually have a conical or bowl-shaped drum with vanes inside. The
efficiency of the mixing operation depends on the details of design, but the discharge
action is always good as all the concrete can be tipped out rapidly and in an
unsegregated mass as soon as the drum is tilted. For this reason, tilting-drum mixers
are preferable for mixes of low workability and for those containing large-size
aggregate.
On the other hand, because of a rather slow rate of discharge from a non-tilting
drum mixer, concrete is sometimes susceptible to segregation. In particular, the
largest size of aggregate may tend to stay in the mixer so that the discharge sometimes
starts as mortar and ends as a collection of coated coarse aggregate particles. Non-
tilting mixers are less frequently used than in the past.
Non-tilting mixers are always charged by means of a loading skip, which is also
used with the larger tilting drum mixers. It is important that the whole charge from the
skip be transferred into the mixer every time, i.e. no sticking must occur. Sometimes,
a shaker mounted on the skip assists in emptying it.
The pan mixer is generally not mobile and is therefore used at a central mixing
plant, at a precast concrete plant, or in a small version in the concrete laboratory. The
mixer consists essentially of a circular pan rotating about its axis, with one or two
stars of paddles rotating about a vertical axis not coincident with the axis of the pan.
Sometimes, the pan is static and the axis of the star travels along a circular path about
the axis of the pan. In either case, the relative movement between the paddles and the
concrete is the same, and concrete in every part of the pan is thoroughly mixed.
Scraper blades prevent mortar sticking to the sides of the pan, and the height of the
paddles can be adjusted so as to prevent a permanent coating of mortar forming on the
bottom of the pan.
Pan mixers offer the possibility of observing the concrete in them, and therefore of
adjusting the mix in some cases. They are particularly efficient with stiff and cohesive
mixes and are, therefore, often used in the manufacture of precast concrete. They are
also suitable, because of the scraping arrangements, for mixing very small quantities
of concrete – hence their use in the laboratory.
It may be relevant to mention that, in drum-type mixers, no scraping of the sides
takes place during mixing so that a certain amount of mortar adheres to the sides of
the drum and stays there until the mixer has been cleaned. It follows that, at the
beginning of concreting, the first mix would leave a large proportion of its mortar
behind, and the discharge would consist largely of coated coarse particles. This initial
batch should not be routinely used. As an alternative, a certain amount of mortar may
be introduced into the mixer prior to the commencement of concreting, a procedure

219
known as ‘buttering’ or priming the mixer. A convenient and simple way is to charge
the mixer with the usual quantities of cement, water and fine aggregate, simply
omitting the coarse aggregate. The mix in excess of that stuck in the mixer can be
used in construction and may in fact be particularly suitable for placing at a cold joint.
The necessity of buttering should not be forgotten in laboratory work.
The nominal size of a mixer is described by the volume of concrete after
compaction (BS 1305 : 1974 (obsolescent)), which may be as low as one-half of the
volume of the unmixed ingredients in a loose state. Mixers are made in a variety of
sizes from 0.04 m3 ( ft3) for laboratory use up to 13 m3 (17 yd3). If the quantity
mixed represents less than one-third of the nominal capacity of the mixer, the
resulting mix may not be uniform, and the operation would, of course, be
uneconomical. Overload not exceeding 10 per cent is generally harmless.
All the mixers considered so far are batch mixers, in that one batch of concrete is
mixed and discharged before any more materials are added. As opposed to this,
a continuous mixer discharges mixed concrete steadily without interruption, being fed
by a continuous volume- or weigh-batching system. The mixer itself consists of a
spiral blade rotated at a relatively high speed in an enclosed, slightly inclined trough.
ASTM C 685-10 prescribes the requirements for concrete made by volumetric
batching and continuous mixing, and ACI 304.6R-914.113 offers a guide for the use of
the relevant equipment. Modern continuous mixers produce concrete of high
uniformity.4.113 Using a continuous-feed mixer, placing, compaction and finishing can
all be achieved within 15 minutes of the introduction of water into the
mix.4.101 Volume-batched mixers are used also with recycled concrete aggregate.4.123
Other mixers should be briefly mentioned. These include revolving-drum truck
mixers, reference to which is made on p. 217. There have also been developed twin-
fin truck mixers with water nozzles distributed within the drum, but no adequate data
on their performance are available.
Specialized mixers are used in shotcreting and for mortar for preplaced aggregate
concrete. In the ‘colloid’ mixer used for the latter, cement and water are formed into
colloidal grout by passage, at a speed of 2000 rev/min, through a narrow gap, and
sand is subsequently added to the grout. The pre-mixing of cement and water allows
better subsequent hydration and, when used for concrete, leads to a higher strength at
a given water/cement ratio than conventional mixing. For instance, at water/cement
ratios of 0.45 to 0.50, a gain in strength of 10 per cent has been observed.4.26 However,
a large amount of heat is generated at very low water/cement ratios.4.64 Moreover, two-
stage mixing undoubtedly represents a higher cost and is likely to be justifiable only
in special cases.
Uniformity of mixing
In any mixer, it is essential that sufficient interchange of materials between different
parts of the chamber takes place, so that uniform concrete is produced. The efficiency
of the mixer can be measured by the variability of the mix discharged into a number
of receptacles without interrupting the flow of concrete. For instance, a rather rigid
test of ASTM C 94-09a (formally applicable only to truck mixers) lays down that
samples of concrete should be taken from about to points of a batch, and the
differences in the properties of the two samples should not exceed any of the
following:

220
 In the United Kingdom, BS 3963 : 1974 (1980) gives a guide to the assessment of
performance of mixers using a specified concrete mix. Tests are made on two samples
from each quarter of a batch. Each sample is subjected to wet analysis and the
following are determined:
water content as percentage of solids to 0.1 per cent
fine aggregate content as percentage of total aggregate to 0.5 per cent
cement as percentage of total aggregate to 0.01 per cent
water/cement ratio to 0.01.
The sampling accuracy is assured by a limit on the average range of pairs. If two
samplers in a pair differ unduly, i.e. their range is an outlier*, that pair of results can
be discarded.
*
See, for instance, J. B. Kennedy and A. M. Neville, Basic Statistical Methods for
Engineers and Scientists, 3rd Edn., 613 pp. (New York and London, Harper and
Row, 1986).
The mixer performance is judged by the average value of the difference between
the highest and the lowest average of pairs of readings for the four samples in each of
three test batches; thus one bad mixing operation does not condemn a mixer. The
maximum acceptable variabilities of the percentages listed earlier are prescribed by an
obsolescent British Standard BS 1305 : 1974 for different maximum aggregate sizes.
Swedish investigations4.115 have shown that the uniformity of the cement content is
the best measure of uniformity of mixing: this is considered to be satisfactory if the
coefficient of variation (see p. 642) does not exceed 6 per cent for mixes with a slump
of at least 20 mm, and 8 per cent for mixes of lower workability.
A method of the determination of the distribution of water or admixture in the mix
by a radioactive tracer has been developed in France.4.116
As far as volume-batched continuous mixers are concerned, the uniformity of
mixing has to be measured by tolerances on the proportions of the mix ingredients.
ASTM C 685-10 prescribes the following percentage values by mass:

221
 The US Army Corps of Engineers Test Method CRD-C 55-924.117 specifies taking
samples from each one-third of a stationary mixer. For mass concrete, the conformity
requirements are given in the Corps of Engineers Guide Specification 03305; these
are similar to those of ASTM C 94-09a, but the allowable range of density is 32
kg/m3 (2 lb/ft3) and, for compressive strength, 10 per cent. These seemingly higher
values are a reflection of the fact that three samples are used, rather than two as in the
test of ASTM C 94-09a.
It can be added that tests on the uniformity of mixing measure not only the
performance of a mixer, but can also be used to assess the effects of a sequence of
charging the mixer.
Mixing time
On a site, there is often a tendency to mix concrete as rapidly as possible, and it is,
therefore, important to know what is the minimum mixing time necessary to produce
a concrete uniform in composition and, as a result, of satisfactory strength. This time
varies with the type of mixer and, strictly speaking, it is not the mixing time but the
number of revolutions of the mixer that is the criterion of adequate mixing. Generally,
about 20 revolutions are sufficient. Because there is an optimum speed of rotation
recommended by the manufacturer of the mixer, the number of revolutions and the
time of mixing are interdependent.
For a given mixer, there exists a relation between mixing time and uniformity of
the mix. Typical data are shown in Fig. 4.16, based on tests by Shalon and
Reinitz,4.22 the variability being represented as the range of strengths of specimens
made from the given mix after a specified mixing time. Figure 4.17 shows the results
of the same tests plotted as the coefficient of variation against mixing time. It is
apparent that mixing for less than 1 to minutes produces an appreciably more
variable concrete, but prolonging the mixing time beyond these values results in no
significant improvement in uniformity.

222
 Fig. 4.16. Relation between compressive strength and mixing time4.22

Fig. 4.17. Relation between the coefficient of variation of strength and mixing
time4.22
The average strength of concrete also increases with an increase in mixing time, as
shown for instance by Abrams’ tests.4.23 The rate of increase falls rapidly beyond about
one minute and is not significant beyond two minutes; sometimes, even a slight

223
decrease in strength has been observed.4.44 Within the first minute, however, the
influence of mixing time on strength is of considerable importance.4.22
As mentioned before, the exact value of the minimum mixing time, which is given
by the mixer manufacturer, varies with the type of mixer and depends also on its size.
What is essential is to ensure uniformity of mixing, which generally can be achieved
by a minimum mixing time of 1 minute for a mixer size of 1 yd3 ( m3) and 15
additional seconds for each additional cubic yard ( m3). This guidance is given both
by ASTM C 94-09a and by ACI 304R-89.4.76 According to ASTM C 94-09a, the
mixing time is reckoned from the time when all the solid materials have been put in
the mixer, and it is also required that all the water has to be added not later than after
one-quarter of the mixing time. ACI 304R-89 reckons the mixing time from the time
when all the ingredients have been discharged into the mixer.
The figures quoted refer to the usual mixers but there are many modern large
mixers which perform satisfactorily with a mixing time of 1 to minutes. In high-
speed pan mixers, the mixing time can be as short as 35 seconds. On the other hand,
when lightweight aggregate is used, the mixing time should be not less than 5 minutes,
sometimes divided into 2 minutes of mixing the aggregate with water, followed by 3
minutes with cement added. In general, the length of mixing time required for
sufficient uniformity of the mix depends on the quality of blending of materials
during charging of the mixer: simultaneous feed is beneficial.
Let us consider now the other extreme – mixing over a long period. Generally,
evaporation of water from the mix takes place, with a consequent decrease in
workability and increase in strength. A secondary effect is that of grinding of the
aggregate, particularly if soft: the grading of the aggregate thus becomes finer, and the
workability lower. The friction effect also produces an increase in the temperature of
the mix.
In the case of air-entrained concrete, prolonged mixing reduces the air content by
about per hour (depending on the type of air-entraining agent), while a delay in
placing without continuous mixing causes a drop in air content by only about per
hour. On the other hand, a decrease in mixing time below 2 or 3 minutes may lead to
inadequate entrainment of air.
Intermittent remixing up to about 3 hours, and in some cases up to 6 hours, is
harmless as far as strength and durability are concerned, but the workability falls off
with time unless loss of moisture from the mixer is prevented. Adding water to restore
workability, known as retempering, will lower the strength of the concrete. This is
considered on p. 218.
No general rules on the order of feeding the ingredients into the mixer can be given
as they depend on the properties of the mix and of the mixer. Generally, a small
amount of water should be fed first, followed by all the solid materials, preferably fed
uniformly and simultaneously into the mixer. If possible, the greater part of the water
should also be fed during the same time, the remainder of the water being added after
the solids. With some drum mixers, however, when a very dry mix is used, it is
necessary to feed first some water and the coarse aggregate, as otherwise its surface
does not become sufficiently wetted. Moreover, if coarse aggregate is totally absent to
begin with, sand or sand and cement become lodged in the head of the mixer and do
not become incorporated in the mix; this is known as head pack. If water or cement

224
are fed too fast, or are too hot, there is a danger of formation of cement balls,
sometimes up to 70 mm (or 3 in.) in diameter. With small laboratory pan mixers and
very stiff mixes, it has been found convenient to feed first the fine aggregate, a part of
the coarse aggregate and cement, then the water, and finally the remainder of the
coarse aggregate so as to break up any nodules of mortar.
Tests on flowing concrete made with a superplasticizer,4.118 have shown the slump
to be highest when cement and fine aggregate are mixed together first, and to be
lowest when cement and water are mixed together first. Mixing all the ingredients
simultaneously resulted in an intermediate slump. Figure 4.18 shows this situation and
shows also that the rate of slump loss was highest when cement and fine aggregate
were mixed together first. The slump loss was lowest when all the materials were
mixed simultaneously. It seems thus that, to minimize slump loss, the conventional
mixing technique is the most beneficial.

Fig. 4.18. Loss of slump with time for concretes with a water/cement ratio of 0.25
and a superplasticizer for different batching sequences: (A) all ingredients
simultaneously; (B) cement and water first; (C) cement and fine aggregate first
(based on ref. 4.118)
In connection with mixing flowing concrete, it is worth noting that visual
judgement of consistency of the mix by the mixer operator is not possible because the
mix simply looks fluid.
Hand mixing

225
There may be rare occasions when small quantities of concrete have to be mixed by
hand and, because in this case uniformity is more difficult to achieve, particular care
and effort are necessary. In order to make sure that the relevant art be not forgotten,
an appropriate procedure will be described.
The aggregate should be spread in a uniform layer on a hard, clean and non-porous
base; cement is then spread over the aggregate, and the dry materials are mixed by
turning over from one end of the tray to the other and ‘cutting’ with a shovel until the
mix appears uniform. Turning three times is usually required. Water is then gradually
added so that neither water by itself nor with cement can escape. The mix is turned
over again, usually three times, until it appears uniform in colour and consistency.
It is obvious that during hand mixing no soil or other extraneous material must be
allowed to become included in the concrete.
Ready-mixed concrete
Ready-mixed concrete used to be treated as a separate topic but, nowadays, with the
vast majority of concrete in many countries originating from a central plant, only
certain special features of ready-mixed concrete will be considered in this section.
Ready-mixed concrete is particularly useful on congested sites or in road
construction where little space for a mixing plant and for extensive aggregate
stockpiles is available, but perhaps the greatest single advantage of ready-mixed
concrete is that it is made under better conditions of control than are normally
possible on any but large construction sites. Control has to be enforced but, since the
central mixing plant operates under near-factory conditions, a really close control of
all operations of production of fresh concrete is possible. Proper care during
transportation of the concrete is also ensured by the use of agitator trucks, but the
placing and compaction remain, of course, the responsibility of the personnel on the
site. The use of ready-mixed concrete is also advantageous when only small quantities
of concrete are required or when concrete is placed only at intervals.
There are two principal categories of ready-mixed concrete. In the first, the mixing
is done at a central plant and the mixed concrete is then transported, usually in an
agitator truck which revolves slowly so as to prevent segregation and undue stiffening
of the mix. Such concrete is known as central-mixed as distinct from the second
category – transit-mixed or truck-mixed concrete. Here, the materials are batched at a
central plant but are mixed in a mixer truck either in transit to the site or immediately
prior to the concrete being discharged. Transit-mixing permits a longer haul and is
less vulnerable in case of delay, but the capacity of a truck used as a mixer is only 63
per cent, or even less, of the drum while for central-mixed concrete it is 80 per cent.
Sometimes, the concrete is partially mixed at a central plant in order to increase the
capacity of the agitator truck. The mixing is completed en route. Such concrete is
known as shrink-mixed concrete but is rarely used. Truck mixers usually have a
capacity of 6 m3 (8 yd3) or 7.5 m3 (10 yd3).
It should be explained that agitating differs from mixing solely by the speed of
rotation of the mixer: the agitating speed is between 2 and 6 rev/min, compared with
the mixing speed of 4 to about 16 rev/min; there is thus some overlap in the
definitions. It may be noted that the speed of mixing affects the rate of stiffening,
while the total number of revolutions controls the uniformity of mixing. Unless the
concrete has been shrink-mixed in the central plant mixer, 70 to 100 revolutions at
mixing speed in the truck mixer are required. An overriding limit of 300

226
revolutions in toto is laid down by ASTM C 94-09a. This is thought to be
unnecessary4.78 unless the aggregate, especially the fine fraction, is soft and liable to
grinding.
If the final part of water is put into the mixer just prior to delivery of the concrete
(as may be desirable in hot weather), ASTM C 94-09a requires 30 additional
revolutions at mixing speed prior to discharge.
The main problem in the production of ready-mixed concrete is maintaining the
workability of the mix right up to the time of placing. Concrete stiffens with time and
the stiffening may also be aggravated by prolonged mixing and by a high temperature.
In the case of transit-mixing, water need not be added till nearer the commencement
of mixing but, according to ASTM C 94-09a, the time during which the cement and
moist aggregate are allowed to remain in contact is limited to 90 minutes; BS 5328 :
3 : 1990 superseded by BS EN 206-1 : 2000 allows 2 hours. The 90-minute limit can
be relaxed by the purchaser of the concrete; there is evidence4.83 that, with the use of
retarders, the time limit can be extended to 3 or even 4 hours, provided the concrete
temperature at delivery is below 32 °C (90 °F).
The United States Bureau of Reclamation provides for an extension of 2 to 6 hours
in the time of contact between cement and wet aggregate in transport prior to mixing.
This requires 5 per cent of additional cement for every hour between these limits; thus
between 5 and 20 per cent additional cement can be required.4.97
Retempering
The loss of slump with time was discussed on p. 205. There are two reasons for this
phenomenon. First, from the instant that cement powder and water come into contact
with one another, chemical reactions of hydration of cement take place. As these
reactions involve fixing of water, less water is left to ‘lubricate’ the movement of
individual particles in the mix. Second, in most ambient conditions, some of the mix
water evaporates into the atmosphere and does so the more rapidly the higher the
temperature and the lower the ambient relative humidity.
We can see, therefore, that, if a specified workability is required at the point of
delivery of the concrete after a certain passage of time, this has to be ensured by the
use of appropriate mix proportions and transport arrangements. Occasionally,
however, delays occur in transport or other mishaps prevent a timely discharge of the
concrete. If, in the meantime, a loss of slump occurs, the question arises as to whether
the slump can be restored by means of addition of water coupled with remixing. Such
an operation is referred to as retempering.
As retempering increases the original water/cement ratio of the mix, it is arguable
that it should not be permitted where the original water/cement ratio was directly or
indirectly specified. This is an appropriate stance under some circumstances but, at
other times, a more flexible and sensible solution may be appropriate as long as the
consequences of retempering are understood and appreciated.
The starting point to be considered is the overall water/cement ratio on the basis
both of the original mix water and the retempering water. There is considerable
evidence4.24,4.45 that not all the retempering water should be counted as part of the free
water for the purpose of calculating the water/cement ratio. The reason for this
behaviour probably lies in the fact that water replacing that lost by evaporation should
not be included in the effective water/cement ratio; only the water replacing that used
in early hydration constitutes part of the effective mix water.

227
 It follows from the above that the relation between strength and the overall free
water/cement ratio for retempered concrete is slightly more advantageous than the
usual ratio between strength and the free water/cement ratio; an example of two such
relations was obtained by Hanayneh and Itani.4.90
Nevertheless, retempering inevitably results in some loss of strength compared
with the original concrete. A loss of 7 to 10 per cent was reported,4.90 but it can be
much higher depending on the amount of retempering water added to the
mix4.28 (see Fig. 4.19). Some empirical relationships have been suggested4.88 but, in
practice, the precise amount of retempering water may not be known, if only because
partial discharge from the mixer had occurred prior to the realization of the slump loss.

Fig. 4.19. Effect of retempering water on the strength of concrete4.28

The amount of water needed to raise the slump by 75 mm (3 in.) depends on the
original slump level, being higher at low slumps; Burg4.89 reported the following (in
litres per cubic metre of concrete):
22 to 32 at a slump of less than 75 mm
14 to 18 at a slump of 75 to 125 mm, and
4 to 9 at a slump of 125 to 150 mm.
Another way of viewing the preceding data is to say that the lower the
water/cement ratio the more retempering water is needed. The amount of water also
rises steeply with an increase in temperature so that at 50 °C (125 °F) it can be about
double that at 30 °C (86 °F).4.121
Pumped concrete
Since this book deals primarily with the properties of concrete, the details of the
means of transporting and placing are not considered; they are dealt with, for example,
in ACI Guide 304R-89.4.76 However, an exception should be made in the case of
pumping of concrete because this means of transportation requires the use of mixes
having special properties.

228
Concrete pumps
The pumping system consists essentially of a hopper into which concrete is
discharged from the mixer, a concrete pump of the type shown in Fig. 4.20 or 4.21,
and pipes through which the concrete is pumped.

Fig. 4.20. Direct-acting concrete pump

Fig. 4.21. Squeeze-type concrete pump

Many pumps are of the direct-acting, horizontal piston type with semi-rotary
valves set so as to permit always the passage of the largest particles of
aggregate being used: there is thus no full closure. Concrete is fed into the pump by
gravity and is also partially sucked in during the suction stroke. The valves open and
close with definite pauses so that concrete moves in a series of impulses but the pipe
always remains full. Modern piston pumps are highly effective.
There exist also portable peristaltic pumps, called squeeze pumps, for use with
small diameter (up to 75 or 100 mm (3 or 4 in.)) pipes; Fig. 4.21 shows such a pump.
Concrete placed in a collecting hopper is fed by rotating blades into a pliable pipe
located in a pumping chamber under vacuum. This ensures that, except when actually
squeezed by a roller, the pipe has a normal (cylindrical) shape so that a continuous
flow of concrete is ensured. Two rotating rollers progressively squeeze the tube and
thus pump the concrete in the suction pipe towards the delivery pipe.
Squeeze pumps move concrete for distances up to 90 m (300 ft) horizontally or 30
m (100 ft) vertically. However, using piston pumps, concrete can be moved up to
about 1000 m (3300 ft) horizontally or 120 m (400 ft) vertically, or to proportionate
combinations of distance and lift. We should note that the ratio of equivalent

229
horizontal and vertical distances varies with the consistency of the mix and with the
velocity of the concrete in the pipe: the greater the velocity the smaller the ratio;4.29 at
0.1 m/s it is 24, but at 0.7 m/s it is only 4.5. Special pumps operating at high pressures
can pump concrete up to 1400 m (4600 ft) horizontally or 430 m (1430 ft)
vertically.4.114 New record values, recently 600 m, continue to be reported.
When bends are used, and these should be kept to a minimum and must never be
sharp, the loss of head should be allowed for in the calculation of the range of
delivery: roughly, each 10° bend is equivalent to a length of pipe up to 1 m.
Pumps of different sizes are available and likewise pipes of various diameters are
used, but the pipe diameter must be at least three times the maximum aggregate size.
It is important to note that oversize in coarse aggregate should not be permitted so as
to avoid blockage at bends.
Using squeeze pumps, an output of 20 m3 (25 yd3) of concrete per hour can be
obtained with 75 mm (3 in.) pipes, but piston pumps with 200 mm (8 in.) pipes can
deliver up to 130 m3 (170 yd3) per hour.
Pumps can be truck- or trailer-mounted and can deliver concrete through a folding
boom. In Japan, a horizontal concrete distributor which automatically controls the
position of the pipe is sometimes used;4.87 this reduces the hard work of controlling the
pipe end during discharge.
Use of pumping
Pumping is economical if it can be used without interruption because, at the beginning
of each period of pumping, the pipes have to be lubricated by mortar (at the rate of
about 0.25 m3 per 100 m (1 yd3 per 1000 ft) of 150 mm (6 in.) diameter pipe) and also
because at the end of the operation a considerable effort is required in cleaning the
pipes. However, alterations to the pipeline system can be made very quickly as special
couplings are used. A short length of flexible hose near the discharge end facilitates
placing but increases the friction loss. Aluminium pipes must not be used because
aluminium reacts with the alkalis in cement and generates hydrogen. This gas
introduces voids in the hardened concrete with a consequent loss of strength, unless
the concrete is placed in a confined space.
The main advantages of pumping concrete are that it can be delivered to points
over a wide area otherwise not easily accessible, with the mixing plant clear of the site;
this is especially valuable on congested sites or in special applications such as tunnel
linings, etc. Pumping delivers the concrete direct from the mixer to the form and so
avoids double handling. Placing can proceed at the rate of the output of the mixer, or
of several mixers, and is not held back by the limitations of the transporting and
placing equipment. A high proportion of ready-mixed concrete is nowadays pumped.
Furthermore, pumped concrete is unsegregated but of course in order to be able to
be pumped the mix must satisfy certain requirements. It might be added that
unsatisfactory concrete cannot be pumped so that any pumped concrete is satisfactory
as far as its properties in the fresh state are concerned. Control of the mix is afforded
by the force required to stir it in the hopper and by the pressure required to pump it.
Requirements for pumped concrete
Concrete which is to be pumped must be well mixed before feeding into the pump,
and sometimes remixing in the hopper by means of a stirrer is carried out. Broadly
speaking, the mix must not be harsh or sticky, too dry or too wet, i.e. its consistency is

230
critical. A slump of between 50 and 150 mm (2 and 6 in.) is generally recommended,
but pumping produces a partial compaction so that at the point of delivery the slump
may be decreased by 10 to 25 mm ( to 1 in.). With a lower water content, the coarse
particles, instead of moving longitudinally in a coherent mass in suspension, would
exert pressure on the walls of the pipe. When the water content is at the correct, or
critical, value, friction develops only at the surface of the pipe and in a thin, 1 to 2.5
mm (0.04 to 0.1 in.), layer of the lubricating mortar. Thus, nearly all the concrete
moves at the same velocity, i.e. by way of plug flow. It is possible that the formation
of the lubricating film is aided by the fact that the dynamic action of the piston is
transmitted to the pipe, but such a film is also caused by steel trowelling of a concrete
surface. To allow for the film in the pipe, a cement content slightly higher than
otherwise would be used is desirable. The magnitude of the friction developed
depends on the consistency of the mix, but there must be no excess water because
segregation would result.
It may be useful to consider the problems of friction and segregation in more
general terms. In a pipe through which a material is pumped, there is a pressure
gradient in the direction of flow due to two effects: head of the material and friction.
This is another way of saying that the material must be capable of transmitting a
sufficient pressure to overcome all resistances in the pipeline. Of all the components
of concrete, it is only water that is pumpable in its natural state, and it is the water,
therefore, that transmits the pressure to the other mix components.
Two types of blockage can occur. In one, water escapes through the mix so that
pressure is not transmitted to the solids, which therefore do not move. This occurs
when the voids in the concrete are not small enough or intricate enough to provide
sufficient internal friction within the mix to overcome the resistance of the pipeline.
Therefore, an adequate amount of closely packed fines is essential to create a ‘blocked
filter’ effect, which allows the water phase to transmit the pressure but not to escape
from the mix. In other words, the pressure at which segregation occurs must be
greater than the pressure needed to pump the concrete.4.30 It should be remembered, of
course, that more fines mean a higher surface area of the solids and therefore a higher
frictional resistance in the pipe.
We can see thus how the second type of blockage can occur. If the fines content is
too high, the friction resistance of the mix can be so large that the pressure exerted by
the piston through the water phase is not sufficient to move the mass of concrete,
which becomes stuck. This type of failure is more common in high strength mixes or
in mixes containing a high proportion of very fine material such as crusher dust or fly
ash, while the segregation failure is more apt to occur in medium or low strength
mixes with irregular or gap grading.
The optimum situation, therefore, is to produce maximum frictional resistance
within the mix with minimum void sizes, and minimum frictional resistance against
the pipe walls with a low surface area of the aggregate. This means that the coarse
aggregate content should be high, but the grading should be such that there is a low
void content so that only little of the very fine material is required to produce the
‘blocked filter’ effect.
The coarse aggregate content should be higher when the sand is fine. For example,
ACI 304.2R4.114 recommends, for aggregate with the maximum size of 20 mm ( in.),
the bulk volume of dry-rodded coarse aggregate of 0.56 to 0.66 when the sand
fineness modulus is 2.40, and 0.50 to 0.60 when it is 3.00. Because the dry-rodded

231
volume (see p. 128) compensates automatically for differences in particle shape, the
values cited are equally appropriate for rounded and angular aggregate. It is important
to remember that the dry-rodded volume is determined as a ratio of the volume of dry-
rodded coarse aggregate to the volume of concrete, on the basis of ASTM Test
Method C 29-09; this ratio is entirely distinct from the mass content of coarse
aggregate per cubic metre of concrete in the actual mix.
Fine aggregate conforming to ASTM C 33-08; but with stricter limits at either of
the permitted extremes, is suitable for use in pumped concrete. Experience has shown
that, for pipes smaller than 125 mm (5 in.), 15 to 30 per cent of the fine aggregate
should be finer than 300 μm (No. 50) sieve, and 5 to 10 per cent should be finer than
150 μm (No. 100) sieve.4.114 Deficiency can be remedied by blending with very fine
material such as crusher dust or fly ash. Crushed-rock fine aggregate can be made
suitable by a small addition of rounded sand.4.114 Grading zones found by experience to
be satisfactory, are shown in Table 4.6.

Table 4.6. Recommended Aggregate Gradings for Pumped Concrete (after ACI
304.2R-91)4.114

British tests4.49 have shown that generally the volumetric cement content (at an
assumed density of 2450 kg/m3) has to be at least equal to the void content of the
aggregate but very fine material other than cement can be included with the latter. The
pattern of the effect of the relation between the cement content and void content on
pumpability is shown in Fig. 4.22.4.50 However, it is only fair to add that theoretical
calculations are not very helpful because the shape of the aggregate particles
influences their void content. Some experimental data are shown in Fig. 4.23: they
indicate that the upper limit of pumpability can be successfully exceeded by very rich
concrete.4.59

232
Fig. 4.22. Pumpability of concrete in relation to cement content and void content
of aggregate4.50

233
 Fig. 4.23. Limits on cement content for aggregates with various void contents
with respect to pumpability of concrete4.59
It may be noted that a sudden rise in pressure caused by a restriction or by a
reduction in the diameter of the pipe may result in segregation of the aggregate which
is left behind as the cement paste moves past the obstacle.4.31
The shape of the aggregate influences the optimum mix proportions for good
pumpability but both rounded and angular coarse aggregate can be used; the latter
requires a higher volume of mortar in the mix.4.114 Natural sands are often particularly
suitable for pumping because of their rounded shape and also because their true
grading is more continuous than with crushed aggregate where, within each size
fraction, there is less variety in size. For both these reasons, the void content is
low.4.49 On the other hand, using combinations of size fractions of crushed aggregate, a
suitable void content can be achieved. However, care is required as many crushed
fines are deficient in the size fraction 300 to 600 μm (No. 50 to No. 30 ASTM) but
have excess of material smaller than 150 μm (No. 100). When using crushed coarse
aggregate, it should be remembered that crusher dust may be present and this should
be taken into account in considering the grading of the fine aggregate. Generally, with

234
crushed coarse aggregate, the fine aggregate content should be increased by about 2
per cent.4.51
Flowing concrete can be pumped but an over-cohesive mix with an increased sand
content should be used.4.119
Any mix selection of concrete to be pumped must be subjected to a test. Although
laboratory pumps have been used to predict the pumpability of concrete,4.79 the
performance of any given mix has to be assessed under the actual site conditions,
including the equipment to be used and the distance through which the concrete is to
be pumped.
Various pumping aids4.67 are available for the purpose of improving cohesion of the
mix through increasing the viscosity of the water and of lubrication of the pipe walls.
The pumping aids are meant to be used in addition to, and not instead of, the selection
of appropriate mix proportions. Entrainment of a limited amount of air, 5 or possibly
6 per cent, is also helpful.4.79 However, excess amount of air would decrease the
pumping efficiency as the air would become compressed.
Pumping lightweight aggregate concrete
In the early days of the development of pumping, there were difficulties with the use
of lightweight aggregate whose surface is not sealed. The reason for this is that, under
pressure, the air in the voids in the aggregate contracts, and water is forced into the
pores with the result that the mix becomes too dry.
A remedy was found in pre-soaking both the coarse and the fine aggregate over a
period of 2 to 3 days or by a very rapid vacuum saturation.4.114 Whereas the absorbed
water does not form part of the free water in the mix (see p. 276) it does affect the
batch proportions by mass. Pumping lightweight concrete vertically up to 320 m
(1050 ft) has been reported.
The use of saturated aggregate may have implications for the resistance of concrete
to freezing and thawing, and a period of several weeks may be necessary prior to
exposure.4.114 However, at very low temperatures, reliance on the waiting period is
inadequate, and use of aggregate with very low absorption, coupled with the use of a
special agent, may be necessary. This agent, added to the mix, enters the pores near
the surface of the aggregate but, when the initial hydration of Portland cement raises
the pH, the viscosity of the agent increases and it forms a high-viscosity layer which
hinders absorption of water due to the pumping pressure.4.82
Shotcrete
This is the name given to mortar or concrete conveyed through a hose and
pneumatically projected at high velocity onto a backup surface. The force of the jet
impacting on the surface compacts the material so that it can support itself without
sagging or sloughing, even on a vertical face or overhead. Other names are also used
for some types of shotcrete, e.g. gunite, but only sprayed concrete is sufficiently
general and is indeed the preferred term in the European Union terminology.
The properties of shotcrete are no different from the properties of conventionally
placed mortar or concrete of similar proportions: it is the method of placing that
bestows on shotcrete significant advantages in many applications. At the same time,
considerable skill and experience are required in the application of shotcrete so that its
quality depends to a large extent on the performance of the operators involved,
especially in control of the actual placing by the nozzle.

235
 Because shotcrete is pneumatically projected on a backup surface and then
gradually built up, only one side of formwork or a substrate is needed. This represents
economy, especially when account is taken of the absence of form ties, etc. On the
other hand, the cement content of shotcrete is high. Also, the necessary equipment and
mode of placing are more expensive than in the case of conventional concrete. For
these reasons, shotcrete is used primarily in certain types of construction: thin, lightly
reinforced sections, such as roofs, especially shell or folded plate, tunnel linings, and
prestressed tanks. Shotcrete is also used in repair of deteriorated concrete, in
stabilizing rock slopes, in encasing steel for fireproofing, and as a thin overlay on
concrete, masonry or steel. If shotcrete is applied to a surface covered by running
water, an accelerator producing flash set, such as washing soda, is used. This
adversely affects strength but makes repair work possible. Admixtures for shotcrete
are specified by BS EN 934-5 : 2007. Generally, shotcrete is applied in a thickness up
to 100 mm (4 in.).
There are two basic processes by which shotcrete is applied. In the dry mix
process (which is the more common of the two, in many parts of the world) cement
and damp aggregate are intimately mixed and fed into a mechanical feeder or gun.
The mixture is then transferred by a feed wheel or distributor (at a known rate) into a
stream of compressed air in a hose, and carried up to the delivery nozzle. The nozzle
is fitted inside with a perforated manifold through which water is introduced under
pressure and intimately mixed with the other ingredients. The mixture is then
projected at high velocity onto the surface to be shotcreted.
The fundamental feature of the wet mix process is that all the ingredients,
including the mixing water, are mixed together to begin with. The mixture is then
introduced into the chamber of the delivery equipment and from there conveyed
pneumatically or by positive displacement. A pump similar to that of Fig. 4.21 can be
used. Compressed air (or in the case of pneumatically conveyed mix, additional air) is
injected at the nozzle, and the material is projected at high velocity onto the surface to
be shotcreted.
Either process can produce excellent shotcrete, but the dry mix process is better
suited for use with porous lightweight aggregate and with flash set accelerators, and is
also capable of greater delivery lengths, as well as of intermittent operation.4.34 The
consistency of the mix can be controlled direct at the nozzle, and higher strengths (up
to 50 MPa (or 7000 psi)) can be readily achieved.4.34 On the other hand, the wet mix
process gives a better control of the quantity of mixing water (which is metered, as
opposed to judgement by the nozzle operator) and of any admixture used. Also, the
wet mix process leads to less dust being produced and possibly to lower rebound. The
process is suitable for large-volume operation.
Because of the high velocity of the impacting jet, not all the shotcrete projected on
a surface remains in position: some material rebounds. This consists of the coarsest
particles in the mix, so that the shotcrete in situ is richer than would be expected from
the mix proportions as batched. This may lead to slightly increased shrinkage. The
rebound is greatest in the initial layers and becomes smaller as a plastic cushion of
shotcrete is built up. Typical percentages of material rebounded are:4.34

236
 The significance of rebound is not so much in the waste of the material as in the
danger from accumulation of rebounded particles in a position where they will
become incorporated in the subsequent layers of shotcrete. This can occur if the
rebound collects in inside corners, at the base of walls, behind reinforcement or
embedded pipes, or on horizontal surfaces. Great care in placing of shotcrete is
therefore necessary, and the use of large reinforcement is undesirable. The latter also
leads to the risk of unfilled pockets behind the obstacle to the jet.
The projected shotcrete has to have a relatively dry consistency so that the material
can support itself in any position; at the same time, the mix has to be wet enough to
achieve compaction without excessive rebound. The usual range of water/cement
ratios is 0.30 to 0.50 for dry mix shotcrete, and 0.40 to 0.55 for the wet
mix.4.34 Recommended aggregate gradings are given in Table 4.7. Curing of shotcrete
is particularly important because the large surface/volume ratio can lead to rapid
drying. Recommended practice is given in ACI 506R-904.34 and in BS EN 14487-2 :
2006.

Table 4.7. Recommended Aggregate Gradings for Shotcrete4.34

Shotcrete exhibits durability comparable with ordinary concrete. The only

reservation concerns the resistance to freezing and thawing, especially in salt
water.4.91 Air entrainment of shotcrete is possible using the wet process, but achieving
an adequately low bubble spacing factor (see p. 547) presents some
difficulties.4.94 However, addition of silica fume (7 to 11 per cent by mass of cement)
leads to adequate resistance to freezing and thawing.4.95 More generally, the addition of
silica fume, in proportion of 10 to 15 per cent of the cement by mass, has been found
to improve the cohesion and adhesion of shotcrete; rebound is reduced.4.32 Such
shotcrete can be put into service at an early age.4.96 For very rapid use in service, dry
process shotcrete can be made using regulated-set cement.4.92 The durability of such
shotcrete is good.
Underwater concrete

237
Placing concrete under water presents some special problems. First of all, washout of
the concrete by the water must be avoided so that placement should take place by
discharge from a steel pipe buried within the already placed, but still mobile, concrete.
The pipe, known as a tremie, has to remain full throughout the concreting operation.
In a way, tremie placing of concrete is similar to pumping but the flow of concrete
takes place under the force of gravity only. Placements to depths of 250 m have been
effected.
Continued discharge of concrete makes it flow laterally, and it is therefore
essential that the concrete mix has appropriate flow characteristics. Moreover, these
characteristics cannot be directly observed. A slump of 150 to 250 mm (6 to 10 in.) is
necessary, depending on the presence of embedded items. Anti-washout admixtures
are effective:4.100 they make the concrete flow when pumped or moved but, when the
concrete is at rest, its viscosity is high.4.98
Relatively rich mixes, containing at least 360 kg/m3 (or 600 lb/yd3) of cementitious
material with about 15 per cent of pozzolanas included to improve the flow of
concrete, have been traditionally recommended.4.76 However, Gerwick and
Holland4.100 pointed out that, in large underwater pours, internal temperatures near the
centre of the concrete can reach 70 to 95 °C (160 to 200 °F) and, on subsequent
cooling, cracking can develop. If the concrete is unreinforced, the cracks can be very
wide. For this reason, Gerwick and Holland4.100 suggest the use of blended cements
containing about 16 per cent Portland cement, 78 per cent coarse-ground blastfurnace
slag, and 6 per cent silica fume. The concrete is pre-cooled to 4 °C (40 °F) prior to
discharge into the tremie. A water/cement ratio of 0.40 to 0.45 is commonly used.
Underwater concreting is a delicate operation which, if incorrectly carried out, can
have undetected but serious consequences; use of experienced personnel is necessary.
Preplaced aggregate concrete
This type of concrete is produced in two stages. In the first operation, uniformly
graded coarse aggregate is placed in the forms; either rounded or crushed aggregate is
suitable. In heavily reinforced areas, compaction should be used. The volume of
coarse aggregate represents about 65 to 70 per cent of the overall volume to be
concreted. The remaining voids are filled with mortar in the second stage.
It is clear that the aggregate in the resulting concrete is of the gap-graded type.
Examples of typical coarse and fine aggregates grading are shown in Tables
4.8 and 4.9, respectively. Optimum packing of the aggregate particles leads to great
theoretical advantages but is not necessarily achieved in practice.

Table 4.8. Typical Gradings of Coarse Aggregate for Preplaced Aggregate

Concrete4.75

238
 Table 4.9. Typical Grading of Fine Aggregate for Preplaced Aggregate
Concrete4.75

The coarse aggregate must be free from dirt and dust because, since these are not
removed in mixing, they would impair bond. Flushing the aggregate in situ might
cause an accumulation of dust in the lower part of the pour which would become a
zone of weakness. The aggregate must be saturated and preferably gently inundated.
The second operation consists of pressure pumping of mortar through slotted pipes,
typically 35 mm (or in.) in diameter and spaced at 2 m (7 ft) centres, starting from
the bottom of the mass, the pipes being gradually withdrawn. Pumping over long
distances is possible. ACI 304.1R-924.75 describes various techniques of mortar placing.
A typical mortar consists of a blend of Portland cement and pozzolana in the ratio
of between 2.5 : 1 and 3.5 : 1, by mass. This cementitious material is mixed with sand
in the ratio of between 1 : 1 and 1 : 1.5, at a water/cement ratio of 0.42 to 0.50. An
intrusion aid is added in order to improve the fluidity of the mortar and to hold the
solid constituents in suspension. The intrusion aid also delays somewhat the stiffening
of the mortar and contains a small amount of aluminium powder, which causes a
slight expansion before setting takes place. Strengths of about 40 MPa (6000 psi) are
usual but higher strengths are also possible.4.75
Preplaced aggregate concrete can be placed in locations not easily accessible by
ordinary concreting techniques; it can also be placed in sections containing a large
number of embedded items that have to be precisely located: this arises, for instance,
in nuclear shields. Likewise, because the coarse and fine aggregate are placed
separately, the danger of segregation of heavy coarse aggregate, especially of steel
aggregate used in nuclear shields, is eliminated. In this case, pozzolana should not be
used because it reduces the density of the concrete and fixes less water.4.63 Because of
the reduced segregation, preplaced aggregate concrete is also suitable for underwater
construction.
The drying shrinkage of preplaced aggregate concrete is lower than that of
ordinary concrete, usually 200 × 10–6 to 400 × 10–6. The reduced shrinkage is due to
the point-to-point contact of the coarse aggregate particles, without a clearance for the
cement paste necessary in ordinary concrete. This contact restrains the amount of
shrinkage that can actually be realized, but occasionally shrinkage cracking can
develop.4.53 Because of the reduced shrinkage, preplaced aggregate concrete is suitable
for the construction of water-retaining and large monolithic structures and for repair
work. The low permeability of preplaced aggregate concrete gives it a high resistance
to freezing and thawing.
Preplaced aggregate concrete may be used in mass construction where the
temperature rise has to be controlled: cooling can be achieved by circulating
refrigerated water round the aggregate and thus chilling it; the water is later displaced

239
by the rising mortar. At the other extreme, in cold weather when frost damage is
feared, steam can be circulated in order to pre-heat the aggregate.
Preplaced aggregate concrete is used also to provide an exposed aggregate finish:
special aggregates are placed against the surfaces and become subsequently exposed
by sandblasting or by acid wash.
Preplaced aggregate concrete appears thus to have many useful features but,
because of numerous practical difficulties, considerable skill and experience in
application of the process are necessary for good results to be obtained.
Vibration of concrete
The purpose of compaction of concrete, known also as consolidation, is to achieve the
highest possible density of the concrete. The oldest means of achieving this is by
ramming or punning, but nowadays this technique is very rarely used. The usual
method of compaction is by vibration.
When concrete is freshly placed in the form, air bubbles can occupy between 5 per
cent (in a mix of high workability) and 20 per cent (in a low-slump concrete) of the
total volume. Vibration has the effect of fluidifying the mortar component of the mix
so that internal friction is reduced and packing of coarse aggregate takes place. It is
with respect to achieving a close configuration of coarse aggregate particles that the
particle shape is of great importance (see p. 115). Continuing vibration expels most of
the remainder of entrapped air, but total absence of entrapped air is not normally
achievable.
Vibration must be applied uniformly to the entire concrete mass as otherwise some
parts of it would not be fully compacted while others might be segregated due to over-
vibration. However, with a sufficiently stiff and well-graded mix, the ill effects of
over-vibration can be largely eliminated. Different vibrators require different
consistency of concrete for most efficient compaction so that the consistency of the
concrete and the characteristics of the available vibrator have to be matched. It is
worth noting that flowing concrete, although it may be self-levelling, does not achieve
full compaction by gravity alone. However, the necessary duration of application of
vibration can be reduced by about one-half compared with ordinary concrete.4.47
Good practical guidance on compaction of concrete is given by Mass4.72 and also in
ACI Guide 309R-87.4.73
Internal vibrators
Of the several types of vibrators, this is the most common one. It consists essentially
of a poker, housing an eccentric shaft driven through a flexible drive from a motor.
The poker is immersed in concrete and thus applies approximately harmonic forces to
it; hence, the alternative names of poker- or immersion vibrator.
The frequency of vibration of a vibrator immersed in concrete varies up to 12 000
cycles of vibration per minute: between 3500 and 5000 has been suggested as a
desirable minimum, with an acceleration of not less than 4g but, more recently,
vibration at 4000 to 7000 cycles has found favour.
The poker is easily moved from place to place, and is applied at 0.5 to 1 m (or 2 to
3 ft) centres for 5 to 30 seconds, depending on the consistency of the mix but, with
some mixes, up to 2 minutes may be required. The relation between the radius of
action of an immersion vibrator and the frequency and amplitude of vibration is
discussed in ACI 309.1R-93.4.74

240
 The actual completion of compaction can be judged by the appearance of the
surface of the concrete, which should be neither honeycombed nor contain an excess
of mortar. Gradual withdrawal of the poker at the rate of about 80 mm/sec (3 in./sec)
is recommended4.17 so that the hole left by the vibrator closes fully without any air
being trapped. The vibrator should be immersed through the entire depth of the
freshly deposited concrete and into the layer below if this is still plastic or can be
brought again to a plastic condition. In this manner, a plane of weakness at the
junction of the two layers can be avoided and monolithic concrete is obtained. With a
lift greater than about 0.5 m (2 ft) the vibrator may not be fully effective in expelling
air from the lower part of the layer. An immersion vibrator will not expel air from the
form boundary so that ‘slicing’ along the form by means of a flat plate on edge is
necessary. The use of absorptive linings to the form is helpful in this respect.
Internal vibrators are comparatively efficient because all the work is done directly
on the concrete, unlike other types of vibrators. Pokers are made in sizes down to 20
mm ( in.) diameter so that they can be used even with heavily reinforced and
relatively inaccessible sections. ACI Guide 309R-874.73 gives useful information on
internal vibrators and on selection of appropriate types. Robot-operated internal
vibrators are available in some countries.
External vibrators
This type of vibrator is rigidly clamped to the formwork resting on an elastic support,
so that both the form and the concrete are vibrated. As a result, a considerable
proportion of the work done is used in vibrating the formwork, which has to be strong
and tight so as to prevent distortion and leakage of grout.
The principle of an external vibrator is the same as that of an internal one, but the
frequency is usually between 3000 and 6000 cycles of vibration per minute, although
some vibrators reach 9000 cycles per minute. Manufacturers’ data have to be
inspected carefully as sometimes the number of ‘impulses’ is quoted, an impulse
being half a cycle. The Bureau of Reclamation4.7 recommends at least 8000 cycles.
The power output varies between 80 and 1100 W.
External vibrators are used for precast or thin in situ sections of such shape or
thickness that an internal vibrator cannot be conveniently used. These vibrators are
effective for concrete sections up to 600 mm (24 in.) thick.4.73
When an external vibrator is used, concrete has to be placed in layers of suitable
depth as air cannot be expelled through too great a thickness of concrete. The position
of the vibrator may have to be changed as concreting progresses if the height is more
than 750 mm (30 in.).4.73
Portable, non-clamped external vibrators may be used at sections not otherwise
accessible, but the range of compaction of this type of vibrator is very limited. One
such vibrator is an electric hammer, sometimes used for compaction of concrete test
specimens.
Vibrating tables
This can be considered as a case of formwork clamped to the vibrator, instead of the
other way round, but the principle of vibrating the concrete and formwork together is
unaltered.
The source of vibration, too, is similar. Generally a rapidly rotating eccentric mass
makes the table vibrate with a circular motion. With two shafts rotating in opposite

241
directions, the horizontal component of vibration can be neutralized so that the table is
subjected to a simple harmonic motion in the vertical direction only. There exist also
some small good quality vibrating tables operated by an electro-magnet fed with
alternating current. The range of frequencies used varies between 50 and about 120
Hz. An acceleration of about 4g to 7g is desirable.4.17 About 1.5g and an amplitude of
40 μm. (0.0015 in.) are believed to be the minima necessary for compaction,4.18 but
with these values a long period of vibration may be necessary. For simple harmonic
motion, the amplitude, a, and the frequency, f, are related by the equation:

acceleration = a(2πf)2.

When concrete sections of different sizes are to be vibrated, and in laboratory use,
a table with a variable amplitude should be used. Variable frequency of vibration is an
added advantage.
In practice, the frequency may rarely be varied during the actual compaction but, at
least theoretically, there are considerable advantages in increasing the frequency and
decreasing amplitude as consolidation progresses. The reason for this lies in the fact
that initially the particles in the mix are far apart and the movement induced has to be
of corresponding magnitude. On the other hand, once partial compaction has taken
place, the use of a higher frequency permits a greater number of adjusting movements
in a given time; a reduced amplitude means that the movement is not too large for the
space available. Vibration at too large an amplitude relative to the inter-particle space
results in the mix being in a constant state of flow so that full compaction is never
achieved. Bresson and Brusin4.71 found that there is an optimum amount of energy of
vibration for every mix, and various combinations of frequency and acceleration will
be satisfactory. However, a prediction of the optimum in terms of mix parameters is
not possible.
A vibrating table provides a reliable means of compaction of precast concrete and
has the advantage of offering uniform treatment.
A variant of the vibrating table is a shock table used sometimes in precast concrete
manufacture. The principle of this process of compaction is rather different from the
high frequency vibration discussed earlier: in a shock table, violent vertical shocks are
imparted at the rate of about 2 to 4 per second. The shocks are produced by a vertical
drop of 3 to 13 mm ( to in.), this being achieved by means of cams. Concrete is
placed in the form in shallow layers while the shock treatment progresses: extremely
good results have been reported but the process is rather specialized and not widely
used.
Other vibrators
Various types of vibrators have been developed for special purposes but only a very
brief mention of these will be made.
A surface vibrator applies vibration through a flat plate direct to the top surface of
the concrete. In this manner, the concrete is restrained in all directions so that the
tendency to segregate is limited; for this reason, a more intense vibration can be used.
An electric hammer can be used as a surface vibrator when fitted with a bit having
a large flat area, say 100 mm by 100 mm (4 in. by 4 in.); one of the main applications
is in compacting test cubes.

242
 A vibrating roller is used for consolidating thin slabs. For road construction
various vibrating screeds and finishers are available; these are discussed in ACI 309R-
87.4.73 A power float is used mainly for granolithic floors in order to bind the
granolithic layer to the main body of the concrete, and is more an aid in finishing than
a means of compaction.
Revibration
It is usual to vibrate concrete immediately after placing so that consolidation is
generally completed before the concrete has stiffened. All the preceding sections refer
to this type of vibration.
It has been mentioned, however, that, in order to ensure good bond between lifts,
the upper part of the underlying lift should be revibrated, provided the lower lift can
still regain a plastic state; settlement cracks and the internal effects of bleeding can
thus be eliminated.
This successful application of revibration raises the question whether revibration
can be more generally used. On the basis of experimental results, it appears that
concrete can be successfully revibrated up to about 4 hours from the time of
mixing4.19 provided the vibrator will sink under its own weight into the
concrete.4.72 Revibration at 1 to 2 hours after placing was found to result in an increase
in the 28-day compressive strength as shown in Fig. 4.24. The comparison is on the
basis of the same total period of vibration, applied either immediately after placing or
in part then, and in part at a specified time later. An increase in strength of
approximately 14 per cent has been reported,4.19 but actual values would depend on the
workability of the mix and on details of the procedure: other investigators have found
increases of 3 to 9 per cent.4.80 In general, the improvement in strength is more
pronounced at earlier ages, and is greatest in concretes liable to high
bleeding4.20 because the trapped water is expelled on revibration. For the same reason,
revibration greatly improves watertightness4.72 and also the bond between concrete and
reinforcement near the top surface of the concrete as trapped bleed water is expelled.
It is possible also that some of the improvement in strength is due to a relief of the
plastic shrinkage stresses around aggregate particles.

Fig. 4.24. Relation between 28-day compressive strength and the time of
revibration4.19

243
 Despite these advantages, revibration is not widely used as it involves an
additional step in the production of concrete, and hence increased cost; also, if applied
too late, revibration can damage the concrete.
Vacuum-dewatered concrete
One solution to the problem of combining a sufficiently high workability with a
minimum water/cement ratio is offered by vacuum-dewatering of freshly placed
concrete.
The procedure is briefly as follows. A mix with a medium workability is placed in
the forms in the usual manner. Because fresh concrete contains a continuous system
of water-filled channels, the application of a vacuum to the surface of the concrete
results in a large amount of water being extracted from a certain depth of the concrete.
In other words, what might be termed ‘water of workability’ is removed when no
longer needed. It may be noted that air bubbles are removed only from the surface as
they do not form a continuous system.
The final water/cement ratio before the concrete sets is thus reduced and, as this
ratio largely controls the strength, vacuum-dewatered concrete has a higher strength
and also a higher density, a lower permeability, and a greater durability, as well as a
higher resistance to abrasion, than would otherwise be obtained. However, some of
the water extracted leaves behind voids, so that the full theoretical advantage of water
removal may not be achieved in practice.4.54 In fact, the increase in strength on vacuum
treatment is proportional to the amount of water removed up to a critical value beyond
which there is no significant increase, so that prolonged vacuum treatment is not
useful. The critical value depends on the thickness of concrete and on the mix
proportions.4.55 Nevertheless, the strength of vacuum-dewatered concrete almost
follows the usual dependence on the final water/cement ratio, as shown in Fig. 4.25.

Fig. 4.25. Relation between the strength of concrete and the calculated
water/cement ratio after vacuum dewatering4.55

244
 The vacuum is applied through porous mats connected to a vacuum pump. The
mats are placed on fine filter pads which prevent the removal of cement together with
the water. The mats can be placed on top of the concrete immediately after screeding,
and can also be incorporated in the inside faces of vertical forms.
Vacuum is created by a vacuum pump; its capacity is governed by the perimeter of
the mat, and not its area. The magnitude of the applied vacuum is usually about 0.08
MPa (11 psi). This vacuum reduces the water content by up to 20 percent. The
reduction is greater nearer to the mat and it is usual to assume the suction to be fully
effective over a depth of 100 to 150 mm (4 to 6 in.) only. The withdrawal of water
produces settlement of the concrete to the extent of about 3 per cent of the depth over
which the suction acts. The rate of withdrawal of water falls off with time, and it has
been found that processing during 15 to 25 minutes is usually most economical. Little
reduction in water content occurs beyond 30 minutes.
Strictly speaking, no suction of water takes place during vacuum-dewatering but
merely a fall of pressure below atmospheric is communicated to the interstitial fluid
of the fresh concrete. This would mean that compaction by atmospheric pressure is
taking place. Thus, the amount of water removed would be equal to the contraction in
the total volume of concrete and no voids would be produced. However, in practice,
some voids are formed and, for the same final water/cement ratio, ordinary concrete
has been found to have a somewhat higher strength than vacuum-dewatered concrete.
This is discernible in Fig. 4.25.
The formation of voids can be prevented if, in addition to vacuum-dewatering,
intermittent vibration is applied; under those circumstances a higher degree of
consolidation is achieved and the amount of water withdrawn can be nearly doubled.
In tests by Garnett,4.21 good results were obtained with vacuum-dewatering for 20
minutes accompanied by vibration between the 4th and 8th minutes, and again
between the 14th and 18th minutes.
Vacuum-dewatering can be used over a fairly wide range of aggregate/cement
ratios and aggregate gradings, but a coarser grading yields more water than a finer one.
Furthermore, some of the finest material is removed by the processing, and fine
materials, such as pozzolanas, should not be incorporated in the mix. The use of a
cement content not exceeding 350 kg/m3 (590 lb/yd3) and of water-reducing
admixtures so that the slump does not exceed 120 mm (5 in.) has been
recommended.4.109
Vacuum-dewatered concrete stiffens very rapidly so that formwork can be
removed within about 30 minutes of casting, even on columns 4.5 m (15 ft) high. This
is of considerable economic value, particularly in a precast concrete factory, as the
forms can be re-used at frequent intervals. Usual curing is essential.
The surface of vacuum-dewatered concrete is entirely free from pitting and the
uppermost 1 mm (0.04 in.) is highly resistant to abrasion. These characteristics are of
special importance in concrete which is to be in contact with water flowing at a high
velocity. Another useful characteristic of vacuum-dewatered concrete is that it bonds
well to old concrete and can, therefore, be used for resurfacing road slabs and in other
repair work. Vacuum treatment thus appears to be a valuable process, which is
extensively used in some countries, especially for slabs and floors.4.54
Permeable formwork

245
A recent development, in some ways similar in concept to vacuum-dewatering, is the
use of permeable formwork. Here, the formwork for vertical surfaces consists of a
polypropylene fabric fixed to plywood backing which contains drain holes. Thus, the
formwork acts as a filter through which air and bleed water escape but the cement is,
for the most part, retained in the body of the concrete, although it is carried towards
the formwork. A local increase in cement content of 20 to 70 kg/m3 (30 to 110 lb/yd3)
has been reported.4.93
In addition to reducing the formwork pressure, the permeable formwork lowers the
water/cement ratio in the surface zone, up to a depth of 20 mm; the reduction varies
steadily from about 0.15 next to the formwork to a negligible amount at a depth of 20
mm.4.99 The effect of the greatly reduced water/cement ratio is to reduce surface
absorption and water permeability of the outer zone of the concrete, which is often
critical from the durability standpoint. It should be noted, however, that 20 mm is less
than the cover to reinforcement under exacting conditions of exposure. The surface
hardness of the concrete is also increased; this improves the resistance of concrete to
cavitation and erosion.
Because much of the surplus mix water escapes in the horizontal direction, the
amount of bleed water at the top surface is reduced. This allows earlier finishing of
the surface but, when ambient conditions are conducive to rapid drying, the absence
of bleeding may lead to plastic shrinkage cracking. Appropriate measures need to be
taken.
The surface produced by permeable formwork is free from bleed streaking and
entrapped air pock-marks, thus enhancing the appearance of the exposed surfaces.
While wet curing following formwork removal is desirable, its absence is less harmful
than is the case with the usual, impermeable formwork.
Analysis of fresh concrete
In considering the ingredients of a concrete mix, we have so far assumed that the
actual proportions correspond to those specified. Modern batching plants can provide
a record of materials in each batch, but this does not include details of aggregate
grading nor sufficient information about the moisture content of the aggregate (see
p. 132). Moreover, if the batch record could be totally relied upon in all cases, there
would be little need for testing the strength of hardened concrete. However, in
practice, mistakes, errors and even deliberate actions can lead to incorrect mix
proportions, and it is sometimes useful to determine the composition of concrete at an
early stage; the two values of greatest interest are the cement content and
water/cement ratio. It is the procedures for determining these values that are referred
to as the analysis of fresh concrete.
ASTM test methods for the determination of the cement and water contents have
been withdrawn. A so-called rapid analysis machine described in
refs 4.57, 4.84 and 4.85 has not proved to be successful.
The U.S. Army4.77 uses a test which relies on chloride titration for determining the
water content and on calcium titration for cement content. The test can be performed
in the field and takes no more than a quarter of an hour. However, the fine part
(smaller than 150 μm (No. 100) sieve) of calcareous aggregate cannot be
distinguished from the cement.

246
 The use of the principle of buoyancy to determine the water/cement ratio of a mix
was used by Naik an Ramme4.86 but it requires the knowledge of the aggregate/cement
ratio in the mix, which may well be uncertain or unreliable.
A pressure-filter method has also been developed in which the material smaller
than 150 μm. (No. 100) sieve is separated out by filtering and pressing dry;4.36 the mass
of cement is taken as the mass of this fraction corrected for aggregate finer than
150 μm (No. 100) sieve in the material as batched. This is a likely source of error.
Separation of cement by flotation has also been developed.4.81
A totally different approach in the determination of cement content of fresh
concrete is based on the separation of cement using a heavy liquid and a
centrifuge.4.38 This has not been very successful, especially when the finest aggregate
particles have a specific gravity not significantly lower than that of cement.
A recent development is to determine the water/cement ratio by the measurement
of electrical resistivity using a probe immersed in fresh concrete.4.124 This approach can
be relied on only with a given mix, the change in resistivity indicating a departure
from the expected w/c.
As far as the determination of the water content in fresh concrete is concerned, this
can be measured by estimating the degree of scattering of thermal neutrons emitted by
a source placed within the bulk of the aggregate or within a sample of the
mix.4.69 Hydrogen is the most important element influencing scattering and retardation
of thermal neutrons and, since hydrogen is almost exclusively bound in water, the
nuclear method can provide a value of the water content with an accuracy of ±0.3 per
cent. The technique also requires the dry density of the aggregate to be taken into
account, and this is calculated from the back-scattering of gamma radiation from a
second source. The complete apparatus comprises gamma and thermal neutron
sources, neutron and scintillation detectors, and associated counters. Calibration is
carried out in situ and is a time-consuming process. Use of microwave oven drying
has been proposed.
We can see that there exist no reliable and practicable procedures for the
measurement of the water/cement ratio of fresh concrete. Indeed, there exists no test
for the composition of fresh concrete that is convenient and reliable enough to be used
as a preplacement acceptance test.
Self-compacting (self-consolidating) concrete
This type of concrete (in American parlance, self-consolidating or SCC) expels
entrapped air without vibration and travels around obstacles such as reinforcement, to
fill all space within the formwork. This is useful with intricate patterns of prestressing
tendons and poorly accessible areas near anchorages. Vibration is noisy and therefore
objectionable to neighbours, especially at night and at weekends. Avoiding this noise
is the second argument for the use of self-consolidating concrete.
There is also a third reason, and that is the health effects of immersion vibrators on
operatives; holding the vibrator damages nerves and blood vessels and causes the so-
called ‘white finger’ or ‘hand vibration’ syndrome. This is obviously socially
undesirable, and in the UK there are regulations on the use of hand-held vibrators. As
yet, self-compacting concrete is not widely used in the UK; in addition to Japan,
Sweden and the Netherlands are leaders in the field. In the USA, the
Precast/Prestressed Concrete Institute (PCI) already has a guide on SCC, and ACI has
produced a very helpful guide, 237R-07.

247
 Interestingly, the impetus for the development of self-compacting concrete came
from the desire to minimize the use of unskilled labour in Japan. There is no doubt
that self-consolidating concrete will become more widespread in the near future, even
for lightweight concrete.
There are three desiderata for the concrete to be classified as self-consolidating:
flowing ability; passing through closely spaced reinforcement; and resistance to
segregation. Various tests for each of the three properties have been proposed but no
comprehensive test method has been standardized. In 2010, there were published five
BS EN standards under the designation BS EN 12350: Part 8: slump-flow; Part 9: V-
funnel test; Part 10: L-box test; Part 11: sieve segregation test; and Part 12: J-ring test.
The means of achieving self-compacting concrete are: the use of more fines
(smaller than 600 μm) than usual; an appropriate viscosity achieved by a controlling
agent; w/c of about 0.4; use of a superplasticizer; a good aggregate shape and texture;
less coarse aggregate than usual (50% by volume of all solids). This may result in
lower aggregate interlock, which is beneficial in shear resistance. Clearly, very good
batching controls are necessary.
Self-compacting concrete is very useful for heavily reinforced members of any
shape and with bottlenecks, both in precast concrete and in situ. The only limitation is
that the top surface must be horizontal. Recent standards are BS EN 206-9 : 2010 and
ASTM C 1712-09.
References
4.1. W. H. GLANVILLE, A. R. COLLINS and D. D. MATTHEWS, The grading of
aggregates and workability of concrete, Road Research Tech. Paper No.
5 (London, HMSO, 1947).
4.2. NATIONAL READY-MIXED CONCRETE ASSOCIATION, Outline and Tables for
Proportioning Normal Weight Concrete, 6 pp. (Silver Spring, Maryland, Oct.
1993).
4.3. ROAD RESEARCH LABORATORY: Design of concrete mixes, D.S.I.R. Road Note No.
4 (London, HMSO, 1950).
4.4. A. R. CUSENS, The measurement of the workability of dry concrete
mixes, Mag. Concr. Res., 8, No. 22, pp. 23–30 (1956).
4.5. T. C. POWERS, Studies of workability of concrete, J. Amer. Concr. Inst., 28,
pp. 419–48 (1932).
4.6. J. W. KELLY and M. POLIVKA, Ball test for field control of concrete
consistency, J. Amer. Concr. Inst., 51, pp. 881–8 (May 1955).
4.7. U.S. BUREAU OF RECLAMATION, Concrete Manual, 8th Edn (Denver, 1975).
4.8. P. KLIEGER, Effect of mixing and curing temperature on concrete strength, J.
Amer. Concr. Inst., 54, pp. 1063–81 (June 1958).
4.9. F. M. LEA, The Chemistry of Cement and Concrete (London, Arnold, 1956).
4.10. T. C. POWERS, The bleeding of portland cement paste, mortar and
concrete, Portl. Cem. Assoc. Bull. No. 2 (Chicago, July 1939).
4.11. H. H. STEINOUR, Further studies of the bleeding of portland cement
paste, Portl. Cem. Assoc. Bull. No. 4 (Chicago, Dec. 1945).
4.12. I. L. TYLER, Uniformity, segregation and bleeding, ASTM Sp. Tech. Publ. No.
169, pp. 37–41 (1956).

248
 4.13. B. C. DOELL, Effect of algae infested water on the strength of concrete, J.
Amer. Concr. Inst., 51, pp. 333–42 (Dec. 1954).
4.14. J. D. DEWAR, Relations between various workability control tests for ready-
mixed concrete, Cement Concr. Assoc. Tech. Report TRA/375 (London, Feb.
1964).
4.15. H. H. STEINOUR, Concrete mix water – how impure can it be? J. Portl. Cem.
Assoc. Research and Development Laboratories, 3, No. 3, pp. 32–50 (Sept.
1960).
4.16. W. J. MCCOY, Water for mixing and curing concrete, ASTM Sp. Tech. Publ.
No. 169, pp. 355–60 (1956).
4.17. JOINT COMMITTEE OF THE I.C.E. AND THE I. STRUCT. E., The Vibration of
Concrete (London, 1956).
4.18. J. KOLEK, The external vibration of concrete, Civil Engineering, 54, No. 633,
pp. 321–5 (London, 1959).
4.19. C. A. VOLLICK, Effects of revibrating concrete, J. Amer. Concr. Inst., 54, pp.
721–32 (March 1958).
4.20. E. N. MATTISON, Delayed screeding of concrete, Constructional Review, 32,
No. 7, p. 30 (Sydney, 1959).
4.21. J. B. GARNETT, The effect of vacuum processing on some properties of
concrete, Cement Concr. Assoc. Tech. Report TRA/326 (London, Oct. 1959).
4.22. R. SHALON and R. C. REINITZ, Mixing time of concrete – technological and
economic aspects, Research Paper No. 7 (Building Research Station,
Technion, Haifa, 1958).
4.23. D. A. ABRAMS, Effect of time of mixing on the strength of concrete, The
Canadian Engineer (25 July, 1 Aug., 8 Aug. 1918, reprinted by Lewis
Institute, Chicago).
4.24. G. C. COOK, Effect of time of haul on strength and consistency of ready-
mixed concrete, J. Amer. Concr. Inst., 39, pp. 413–26 (April 1943).
4.25. D. A. ABRAMS, Tests of impure waters for mixing concrete, J. Amer. Concr.
Inst., 20, pp. 442–86 (1924).
4.26. W. JURECKA, Neuere Entwicklungen und Entwicklungstendenzen von Beton-
mischern und Mischanlagen, Österreichischer Ingenieur-Zeitschrift, 10, No.
2, pp. 27–43 (1967).
4.27. K. THOMAS and W. E. A. LISK, Effect of sea water from tropical areas on
setting times of cements, Materials and Structures, 3, No. 14, pp. 101–5
(1970).
4.28. R. C. MEININGER, Study of ASTM limits on delivery time, Nat. Ready-mixed
Concr. Assoc. Publ. No. 131, 17 pp. (Washington DC, Feb. 1969).
4.29. R. WEBER, Rohrförderung von Beton, Düsseldorf Beton-Verlag GmbH
(1963), The transport of concrete by pipeline (London, Cement and
Concrete Assoc. Translation No. 129, 1968).
4.30. E. KEMPSTER, Pumpable concrete, Current Paper 26/69, 8 pp. (Building
Research Station, Garston, 1968).
4.31. E. KEMPSTER, Pumpability of mortars, Contract Journal, 217, pp. 28–30 (4
May 1967).

249
 4.32. T. C. HOLLAND and M. D. LUTHER, Improving concrete quality with silica fume,
in Concrete and Concrete Construction, Lewis H. Tuthill Int. Symposium,
ACI SP-104, pp. 107–22 (Detroit, Michigan, 1987).
4.33. W. J. MCCOY, Mixing and curing water for concrete, ASTM Sp. Tech. Publ.
No. 169B, pp. 765–73 (1978).
4.34. ACI 506.R-90, Guide to shotcrete, ACI Manual of Concrete Practice, Part 5:
Masonry, Precast Concrete, Special Processes, 41 pp. (Detroit, Michigan,
1994).
4.35. BUILDING RESEARCH STATION, Analysis of water encountered in
construction, Digest No. 90 (HMSO, London, July 1956).
4.36. R. BAVELJA, A rapid method for the wet analysis of fresh
concrete, Concrete, 4, No. 9, pp. 351–3 (London, 1970).
4.37. F. P. NICHOLS, Manufactured sand and crushed stone in portland cement
concrete, Concrete International, 4, No. 8, pp. 56–63 (1982).
4.38. W. G. HIME and R. A. WILLIS, A method for the determination of the cement
content of plastic concrete, ASTM Bull. No. 209, pp. 37–43 (Oct. 1955).
4.39. A. MOR and D. RAVINA, The DIN flow table, Concrete International, 8, No.
12, pp. 53–6 (1986).
4.40. O. Z. CEBECI and A. M. SAATCI, Domestic sewage as mixing water in
concrete, ACI Materials Journal, 86, No. 5, pp. 503–6 (1989).
4.41. K. W. NASSER, New and simple tester for slump of concrete, J. Amer. Concr.
Inst., 73, pp. 561–5 (Oct. 1976).
4.42. K. W. NASSER and N. M. REZK, New probe for testing workability and
compaction of fresh concrete, J. Amer. Concr. Inst., 69, pp. 270–5 (May
1972).
4.43. G. H. TATTERSALL, Workability and Quality Control of Concrete, 262 pp. (E &
FN Spon, London, 1991).
4.44. E. NEUBARTH, Einfluss einer Unterschreitung der Mindestmischdauer auf die
Betondruckfestigkeit, Beton, 20, No. 12, pp. 537–8 (1970).
4.45. F. W. BEAUFAIT and P. G. HOADLEY, Mix time and retempering studies on
ready-mixed concrete, J. Amer. Concr. Inst., 70, pp. 810–13 (Dec. 1973).
4.46. ACI 116R-90, Cement and concrete terminology, ACI Manual of Concrete
Practice, Part 1: Materials and General Properties of Concrete, 68 pp.
(Detroit, Michigan, 1994).
4.47. L. FORSSBLAD, Need for consolidation of superplasticized concrete mixes,
in Consolidation of Concrete, Ed. S. H. Gebler, ACI SP-96, pp. 19–37
(Detroit, Michigan, 1987).
4.48. G. HILL BETANCOURT, Admixtures, workability, vibration and
segregation, Materials and Structures, 21, No. 124, pp. 286–8 (1988).
4.49. DEPARTMENT OF THE ENVIRONMENT, Guide to Concrete Pumping, 49 pp. (HMSO,
London, 1972).
4.50. A. JOHANSSON and K. TUUTTI, Pumped concrete and pumping of concrete, CBI
Research Reports, 10: 76 (Swedish Cement and Concrete Research Inst.,
1976).

250
 4.51. J. R. ILLINGWORTH, Concrete pumps – planning considerations, Concrete, 5,
No. 12, p. 387 (London, 1969).
4.52. M. MITTELACHER, Re-evaluating the slump test, Concrete International, 14,
No. 10, pp. 53–6 (1992).
4.53. CUR REPORT, Underwater concrete, Heron, 19, No. 3, 52 pp. (Delft, 1973).
4.54. R. MALINOWSKI and H. WENANDER, Factors determining characteristics and
composition of vacuum dewatered concrete, J. Amer. Concr. Inst., 72, pp.
98–101 (March 1975).
4.55. G. DAHL, Vacuum concrete, CBI Reports, 7: 75, Part 1, 10 pp. (Swedish
Cement and Concrete Research Inst., 1975).
4.56. P. BARTOS, Fresh Concrete, 292 pp. (Amsterdam, Elsevier, 1992).
4.57. I. COOPER and P. BARBER, Field Investigation of the Accuracy of the
Determination of the Cement Content of Fresh Concrete by Use of the C. &
C.A. Rapid Analysis Machine (R.A.M.), 19 pp. (British Ready Mixed
Concrete Assoc., Dec. 1976).
4.58. R. HARD and N. PETERSONS, Workability of concrete – a testing method, CBI
Reports, 2: 76, pp. 2–12 (Swedish Cement and Concrete Research Inst.,
1976).
4.59. A. JOHANSSON, N. PETERSONS and K. TUUTTI, Pumpable concrete and concrete
pumping, CBI Reports, 2: 76, pp. 13–28 (Swedish Cement and Concrete
Research Inst., 1976).
4.60. L. M. MEYER and W. F. PERENCHIO, Theory of Concrete Slump Loss Related to
Use of Chemical Admixtures, PCA Research and Developmen Bulletin
RD069.01T, 8 pp. (Skokie, Illinois, 1980).
4.61. V. DODSON, Concrete Admixtures, 211 pp. (New York, Van Nostrand
Reinhold, 1990).
4.62. V. S. RAMACHANDRAN, Ed., Concrete Admixtures Handbook’
Properties, Science and Technology, 626 pp. (New Jersey, Noyes
Publications, 1984).
4.63. B. A. LAMBERTON, Preplaced aggregate concrete, ASTM Sp. Tech. Publ. No.
169B, pp. 528–38 (1978).
4.64. M. L. BROWN, H. M. JENNINGS and W. B. LEDBETTER, On the generation of heat
during the mixing of cement pastes, Cement and Concrete Research, 20, No.
3, pp. 471–4 (1990).
4.65. T. SOSHIRODA, Effects of bleeding and segregation on the internal structure of
hardened concrete, in Properties of Fresh Concrete, Ed. H.-J. Wierig, pp.
253–60 (London, Chapman and Hall, 1990).
4.66. P. SCHIESSL and R. SCHMIDT, Bleeding of concrete, in Properties of Fresh
Concrete, Ed. H.-J. Wierig, pp. 24–32 (London, Chapman and Hall, 1990).
4.67. ACI 212.3R-91, Chemical admixtures for concrete, ACI Manual of Concrete
Practice, Part 1: Materials and General Properties of Concrete, 31 pp.
(Detroit, Michigan, 1994).
4.68. Y. YAMAMOTO and S. KOBAYASHI, Effect of temperature on the properties of
superplasticized concrete, ACI Journal, 83, No. 1, pp. 80–8 (1986).

251
 4.69. J.-P. BARON, Détermination de la teneur en eau des granulats et du béton frais
par méthode neutronique, Rapport de Recherche LPC No. 72, 56 pp.
(Laboratoire Central des Ponts et Chaussées, Nov. 1977).
4.70. ACI 211.3-75, Revised 1987, Reapproved 1992, Standard practice for
selecting proportions for no-slump concrete, ACI Manual of Concrete
Practice, Part 1: Materials and General Properties of Concrete, 19 pp.
(Detroit, Michigan, 1994).
4.71. J. BRESSON and M. BRUSIN, Etude de l’influence des paramètres de la vibration
sur le comportement des bétons, CERIB Publication No. 32, 23 pp. (Centre
d’Etudes et de Recherche de l’Industrie du Béton Manufacturé, 1977).
4.72. G. R. MASS, Consolidation of concrete, in Concrete and Concrete
Construction, Lewis H. Tuthill Symposium, ACI SP 104-10, pp. 185–203
(Detroit, Michigan, 1987).
4.73. ACI 309R-87, Guide for consolidation of concrete, ACI Manual of Concrete
Practice, Part 2: Construction Practices and Inspection Pavements, 19 pp.
(Detroit, Michigan, 1994).
4.74. ACI 309.1 R-93, Behavior of fresh concrete during vibration, ACI Manual of
Concrete Practice, Part 2: Construction Practices and Inspection
Pavements, 19 pp. (Detroit, Michigan, 1994).
4.75. ACI 304.1R-92, Guide for the use of preplaced aggregate concrete for
structural and mass concrete applications, ACI Manual of Concrete
Practice, Part 2: Construction Practices and Inspection Pavements, 19 pp.
(Detroit, Michigan, 1994).
4.76. ACI 304.R-89, Guide for measuring, mixing, transporting, and placing
concrete, ACI Manual of Concrete Practice, Part 2: Construction Practices
and Inspection Pavements, 49 pp. (Detroit, Michigan, 1994).
4.77. P. A. HOWDYSHELL, Revised operations guide for a chemical technique to
determine water and cement content of fresh concrete, Technical Report M-
212, 36 pp. (US Army Construction Engineering Research Laboratory, April
1977).
4.78. R. D. GAYNOR, Ready-mixed concrete, in Significance of Tests and Properties
of Concrete and Concrete-Making Materials, Eds P. Klieger and J. F.
Lamond, ASTM Sp. Tech. Publ. No. 169C, pp. 511–21 (Philadelphia, Pa,
1994).
4.79. J. F. BEST and R. O. LANE, Testing for optimum pumpability of
concrete, Concrete International, 2, No. 10, pp. 9–17 (1980).
4.80. C. MACINNIS and P. W. KOSTENIUK, Effectiveness of revibration and high-
speed slurry mixing for producing high-strength concrete. J. Amer. Concr.
Inst., 76, pp. 1255–65 (Dec. 1979).
4.81. E. NÄGELE and H. K. HILSDORF, A new method for cement content
determination of fresh concrete, Cement and Concrete Research, 10, No. 1,
pp. 23–34 (1980).
4.82. T. YONEZAWA et al., Pumping of lightweight concrete using non-presoaked
lightweight aggregate, Takenaka Technical Report, No. 39, pp. 119–32
(May 1988).

252
 4.83. F. A. KOZELISKI, Extended mix time concrete, Concrete International, 11, No.
11, pp. 22–6 (1989).
4.84. A. C. EDWARDS and G. D. GOODSALL, Analysis of fresh concrete: repeatability
and reproducibility by the rapid analysis machine, Transport and Road
Research Laboratory Supplementary Report 714, 22 pp. (Crowthorne, U.K.
1982).
4.85. R. K. DHIR, J. G. I. MUNDAY and N. Y. Ho, Analysis of fresh concrete:
determination of cement content by the rapid analysis machine, Mag. Concr.
Res., 34, No. 119, pp. 59–73 (1982).
4.86. T. R. NAIK and B. W. RAMME, Determination of the water–cement ratio of
concrete by the buoyancy principle, ACI Materials Journal, 86, No. 1, pp.
3–9 (1989).
4.87. Y. KAJIOKA and T. FUJIMORI, Automating concrete work in Japan, Concrete
International, 12, No. 6, pp. 27–32 (1990).
4.88. K. H. CHEONG and S. C. LEE, Strength of retempered concrete. ACI Materials
Journal, 90, No. 3, pp. 203–6 (1993).
4.89. G. R. U. BURG, Slump loss, air loss, and field performance of concrete, ACI
Journal, 80, No. 4, pp. 332–9 (1983).
4.90. B. J. HANAYNEH and R. Y. ITANI, Effect of retempering on the engineering
properties of superplasticized concrete, Materials and Structures, 22, No.
129, pp. 212–19 (1989).
4.91. G. W. SEEGEBRECHT, A. LITVIN and S. H. GEBLER, Durability of dry-mix
shotcrete Concrete International, 11, No. 10, pp. 47–50 (1989).
4.92. S. H. GEBLER, Durability of dry-mix shotcrete containining regulated-set
cement Concrete International, 11, No. 10, pp. 56–8 (1989).
4.93. Y. KASAI et al., Comparison of cement contents in concrete surface prepared
in permeable form and conventional form, CAJ Review, pp. 298–301 (1988).
4.94. D. R. MORGAN, Freeze–thaw durability of shotcrete, Concrete
International, 11, No. 8, pp. 86–93 (1989).
4.95. I. L. GLASSGOLD, Shotcrete durability: an evaluation, Concrete
International, 11, No. 8, pp. 78–85 (1989).
4.96. D. R. MORGAN, Dry-mix silica fume shotcrete in Western Canada, Concrete
International, 10, No. 1, pp. 24–32 (1988).
4.97. U.S. BUREAU OF RECLAMATION, Specifications for ready-mixed concrete, 4094-
92, Concrete Manual, Part 2, 9th Edn, pp. 143–59 (Denver, Colorado, 1992).
4.98. K. H. KHAYAT, B. C. GERWICK JNR and W. T. HESTER, Self-levelling and stiff
consolidated concretes for casting high-performance flat slabs in
water, Concrete International, 15, No. 8, pp. 36–43 (1993).
4.99. W. F. PRICE and S. J. WIDDOWS, The effects of permeable formwork on the
surface properties of concrete, Mag. Concr. Res., 43, No. 155, pp. 93–104
(1991).
4.100. B. C. GERWICK JNR and T. C. HOLLAND, Underwater concreting: advancing the
state of the art for structural tremie concrete, in Concrete and Concrete
Construction, ACI SP-104, pp. 123–43 (Detroit, Michigan, 1987).

253
 4.101. N. A. CUMMING and P. T. SEABROOK, Quality assurance program for volume-
batched high-strength concrete, Concrete International, 10, No. 8, pp. 28–32
(1988).
4.102. A. A. AL-MANASEER, M. D. HAUG and K. W. NASSER, Compressive strength of
concrete containing fly ash, brine, and admixtures, ACI Materials
Journal, 85, No. 2, pp. 109–16 (1988).
4.103. H. Y. GHORAB, M. S. HILAL and E. A. KISHAR, Effect of mixing and curing
waters on the behaviour of cement pastes and concrete. Part I:
microstructure of cement pastes, Cement and Concrete Research, 19, No. 6,
pp. 868–78 (1989).
4.104. O. H. WALLEVIK and O. E. GJØRV, Modification of the two-point workability
apparatus, Mag. Concr. Res., 42, No. 152, pp. 135–42 (1990).
4.105. D. L. KANTRO, Influence of water-reducing admixtures on properties of
cement paste – a miniature slump test, Research and Development Bulletin,
RD079.01T, Portland Cement Assn, 8 pp. (1981).
4.106. A. A. AL-MANASEER, K. W. NASSER and M. D. HAUG, Consistency and
workability of flowing concrete, Concrete International, 11, No. 10, pp. 40–
4 (1989).
4.107. J. MURATA and H. KIKUKAWA, Viscosity equation for fresh concrete, ACI
Materials Journal, 89, No. 3, pp. 230–7 (1992).
4.108. B. ERLIN and W. G. HIME, Concrete slump loss and field examples of
placement problems, Concrete International, 1, No. 1, pp. 48–51 (1979).
4.109. S. S. PICKARD, Vacuum-dewatered concrete, Concrete International, 3, No.
11, pp. 49–55 (1981).
4.110. S. SMEPLASS, Applicability of the Bingham model to high strength concrete,
RILEM International Workshop on Special Concretes: Workability and
Mixing, pp. 179–85 (University of Paisley, Scotland, 1993).
4.111. J. M. SHILSTONE SNR, Interpreting the slump test, Concrete International, 10,
No. 11, pp. 68–70 (1988).
4.112. B. SCHWAMBORN, Über das Bluten von Frischbeton, in Proceedings of a
colloquium, Frischmörtel, Zementleim, Frischbeton, University of Hanover,
Publication No. 55, pp. 283–97 (Oct. 1987).
4.113. ACI 304.6R-91, Guide for the use of volumetric-measuring and continuous-
mixing concrete equipment, ACI Manual of Concrete Practice, Part 2:
Construction Practices and Inspection Pavements, 14 pp. (Detroit, Michigan,
1994).
4.114. ACI 304.2R-91, Placing concrete by pumping methods, ACI Manual of
Concrete Practice, Part 2: Construction Practices and Inspection
Pavements, 17 pp. (Detroit, Michigan, 1994).
4.115. Ö. PETERSSON, Swedish method to measure the effectiveness of concrete
mixers, RILEM International Workshop on Special Concretes: Workability
and Mixing, pp. 19–27 (University of Paisley, Scotland, 1993).
4.116. R. BOUSSION and Y. CHARONAT, Les bétonnières portées sont-elles des
mélangeurs?, Bulletin Liaison Laboratoires des Ponts et Chaussées, 149, pp.
75–81 (May–June, 1987).

254
 4.117. U.S. ARMY CORPS of ENGINEERS, Standard test method for within-batch
uniformity of freshly mixed concrete, CRD-C 55–92, Handbook for
Concrete and Cement, 6 pp. (Vicksburg, Miss., Sept. 1992).
4.118. M. KAKIZAKI et al., Effect of mixing method on mechanical properties and
pore structure of ultra high-strength concrete, Katri Report, No. 90, 19 pp.
(Kajima Corporation, Tokyo, 1992) [and also in ACI SP-132, Detroit,
Michigan, 1992].
4.119. P. C. HEWLETT, Ed., Cement Admixtures, Use and Applications, 2nd Edn, for
The Cement Admixtures Association, 166 pp. (Harlow, Longman, 1988).
4.120. E. BIELAK, Testing of cement, cement paste and concrete, including bleeding.
Part 1: laboratory test methods, in Properties of Fresh Concrete, Ed. H.-J.
Wierig, pp. 154–66 (London, Chapman and Hall, 1990).
4.121. S. SASIADEK and M. SLIWINSKI, Means of prolongation of workability of fresh
concrete in hot climate conditions, in Properties of Fresh Concrete, Ed. H.-J.
Wierig, Proc. RILEM Colloquium, Hanover, pp. 109–15 (Cambridge,
University Press, 1990).
4.122. A. NEVILLE, Neville on Concrete: An Examination of Issues in Concrete
Practice, 2nd Edition (Book Surge, LLC, and www.amazon.com, 2006).
4.123. I. BRADBURY, Volumetric mixing with recycled aggregates, Concrete, 44, No.
11, pp. 40–41 (2010).
4.124. M. MANZIO et al., Instantaneous in-situ determination of water-cement ratio
of fresh concrete, ACI Materials Journal, 107, No. 6, pp. 586–92 (2010).

255
Chapter 5. Admixtures

The early chapters described the properties of Portland cement and of a wide range of
cementitious materials, as well as of the aggregate used in making concrete, together
with a discussion of the influence of these materials and their combinations on the
properties of fresh concrete. To a lesser extent, the influence on the properties of
hardened concrete was also discussed but, before the latter topic is considered more
fully, it is useful to review one more ingredient of the concrete mix: admixtures.
While admixtures, unlike cement, aggregate and water, are not an essential
component of the concrete mix, they are an important and increasingly widespread
component: in many countries, a mix which contains no admixtures is nowadays an
exception.
Benefits of admixtures
The reason for the large growth in the use of admixtures is that they are capable of
imparting considerable physical and economic benefits with respect to concrete.
These benefits include the use of concrete under circumstances where previously
there existed considerable, or even insuperable, difficulties. They also make possible
the use of a wider range of ingredients in the mix.
Admixtures, although not always cheap, do not necessarily represent additional
expenditure because their use can result in concomitant savings, for example, in the
cost of labour required to effect compaction, in the cement content which would
otherwise be necessary, or in improving durability without the use of additional
measures.
It should be stressed that, while properly used admixtures are beneficial to concrete,
they are no remedy for poor quality mix ingredients, for use of incorrect mix
proportions, or for poor workmanship in transporting, placing and compaction.
Types of admixtures
An admixture can be defined as a chemical product which, except in special cases, is
added to the concrete mix in quantities no larger than 5 per cent by mass of cement
during mixing or during an additional mixing operation prior to the placing of
concrete, for the purpose of achieving a specific modification, or modifications, to the
normal properties of concrete.
Admixtures may be organic or inorganic in composition but their chemical
character, as distinct from mineral, is their essential feature. Indeed, in American
nomenclature, they are called chemical admixtures but in this book such a
qualification is superfluous because the mineral products incorporated in the mix,
almost invariably in excess of 5 per cent of the mass of cement, are referred to as
cementitious materials or as additives.
Admixtures are commonly classified by their function in concrete but often they
exhibit some additional action. The classification of ASTM C 494-10 is as follows:
Type A Water-reducing
Type B Retarding
Type C Accelerating
Type D Water-reducing and retarding

256
 Type E Water-reducing and accelerating
Type F High-range water-reducing or superplasticizing, and
Type G High-range water-reducing and retarding, or superplasticizing and
retarding
The British Standard for admixtures is BS EN 934-2 : 2009. Also relevant are BS
EN 480 – numerous parts.
In practice, admixtures are marketed as proprietary products, and promotional
literature sometimes includes claims of varied and wide-ranging benefits. While these
may be true, some of the benefits occur only indirectly as a consequence of particular
circumstances so that it is important to understand the specific effects of admixtures
before they are used. Moreover, as ASTM C 494-10 points out, the specific effects
produced may vary with the properties and proportions of the other ingredients of the
mix.
Admixtures may be used in solid or liquid state. The latter is usual because a liquid
can be more rapidly dispersed in a uniform manner during mixing of concrete.
Properly calibrated dispensers are used, the admixture being discharged into the
mixing water, or separately in dilute form but simultaneously with the mixing water,
usually during the latter part of the water feed. Superplasticizers are subject to special
methods of incorporation into the mix.
The dosages of the various admixtures, usually expressed as a percentage of the
mass of cement in the mix, are recommended by the manufacturers but they are often
varied according to circumstances.
The effectiveness of any admixture may vary depending on its dosage in the
concrete and also on the constituents of the mix, especially the properties of the
cement. With some admixtures, the relevant dosage is the solids content and not the
total mass of the admixture in liquid form. However, as far as the water content of the
mix is concerned, the total volume of the liquid admixtures should be counted in but
the solids content of superplasticizers should be excluded.
It is important that the effect of any admixture should not be highly sensitive to
small variations in its dosage as such variations can occur accidentally during the
production of concrete. The effect of many admixtures is influenced by temperature;
for this reason their performance at extreme temperatures should be ascertained prior
to use.
Admixtures should, generally speaking, not be allowed to come into contact with
skin or eyes.
In addition to the chemical admixtures which will be discussed in this chapter,
there exist also air-entraining agents which are considered in Chapter 11.
Accelerating admixtures
For brevity, these ASTM Type C admixtures will be referred to as accelerators. Their
function is primarily to accelerate the early strength development of concrete, that is
hardening (see p. 19), although they may also coincidentally accelerate the setting of
concrete. If a distinction between the two actions is required, it may be useful to refer
to set-accelerating properties.
Accelerators may be used when concrete is to be placed at low temperatures, say 2
to 4 °C (35 to 40 °F), in the manufacture of precast concrete (where a rapid removal
of formwork is desirable) or in urgent repair work. Other benefits of using an

257
accelerator are that it allows earlier finishing of the concrete surface and application
of insulation for protection, and also putting the structure into service earlier.
Conversely, at high temperatures, accelerators may result in too high a rate of
development of heat of hydration and in shrinkage cracking.5.4
While accelerators are often used at very low temperatures, they are not anti-
freezing agents; they depress the freezing point of concrete by no more than 2 °C (or
about 3.5 °F), so that the usual anti-freezing precautions should always be taken (see
p. 405). Special anti-freezing agents are being developed5.8,5.9 but are still not fully
proven.
The most common accelerator used over many decades was calcium chloride.
Calcium chloride is effective in accelerating the hydration of the calcium silicates,
mainly C3S, possibly by a slight change in the alkalinity of the pore water or as a
catalyst in the reactions of hydration. Although the mechanism of its action is even
now imperfectly understood, there is no doubt that calcium chloride is an effective
and cheap accelerator but it has one serious defect: the presence of chloride ions in the
vicinity of steel reinforcement or other embedded steel is highly conducive to
corrosion; this topic is discussed in Chapter 11.
Although the reactions of corrosion take place only in the presence of water and
oxygen, the risks attendant on the presence of chloride ions in concrete containing
steel are such that calcium chloride should never be incorporated into reinforced
concrete; in prestressed concrete, the risks are even higher. In consequence, various
standards and codes prohibit the use of calcium chloride in concrete containing
embedded steel or aluminium. Moreover, even in plain concrete, when its durability
may be impaired by outside agencies, the use of calcium chloride may be inadvisable.
For instance, the resistance of cement to sulfate attack is reduced by the addition of
CaCl2 to lean mixes, and the risk of an alkali-aggregate reaction, when the aggregate
is reactive, is increased.5.24 However, when this reaction is effectively controlled by the
use of low-alkali cement and the addition of pozzolanas, the effect of CaCl2 is very
small. Another undesirable feature of the addition of CaCl2 is that it increases the
drying shrinkage usually by about 10 to 15 per cent, sometimes even more,5.24 and
possibly increases also the creep.
Although the addition of CaCl2 reduces the danger of frost attack during the first
few days after placing, the resistance of air-entrained concrete to freezing and thawing
at later ages is adversely affected. Some indication of this is given in Fig. 5.1.

258
 Fig. 5.1. Resistance to freezing and thawing of concrete cured moist at 4 °C
(40 °F) for different contents of CaCI25.24
On the credit side, CaCl2 has been found to raise the resistance of concrete to
erosion and abrasion, and this improvement persists at all ages.5.24 When plain concrete
is steam cured, CaCl2 increases the strength of concrete and permits the use of a more
rapid temperature rise during the curing cycle5.25 (see p. 370).
The action of sodium chloride is similar to that of calcium chloride but is of lower
intensity. The effects of NaCl are also more variable and a depression in the heat of
hydration, with a consequent loss of strength at 7 days and later, has been observed.
For this reason, the use of NaCl is definitely undesirable. Barium chloride has been
suggested but it acts as an accelerator only under warm conditions.5.44
Some researchers express the view that the use of calcium chloride does not
contribute significantly to the corrosion of steel reinforcement if concrete is well
proportioned and well compacted, and if the cover to the reinforcement is
adequate.5.53 Unfortunately, on site, such perfection may, from time to time, not be
achieved, and the risk of using calcium chloride greatly outweighs its benefits.
Moreover, experience has shown that, under extreme conditions of exposure existing
in some countries, only high performance concrete would protect the reinforcement
from corrosion (see Chapter 13).
Because of this concern about the corrosion of reinforcement, the use, properties
and effects of calcium chloride will not be discussed further in the present book. This
concern has led to the search for chloride-free accelerators. No single accelerator has
become widely accepted but a description of those which can be used may be of value.
Calcium nitrite and calcium nitrate are possible accelerators; the former also
appears to be a corrosion inhibitor.5.1 Calcium formate and sodium formate are also
possibilities, although the latter would introduce sodium into the mix, and this alkali

259
is known to influence hydration and also has a potential reaction with some
aggregates (see p. 145).
Calcium formate is effective only when used with cements which have a ratio of
C3A to SO3 of at least 4 and have a low SO3 content; cements produced using coal with
a relatively high sulfur content do not satisfy this requirement.5.7 For this reason, trial
mixes involving any given cement should be made. It may be noted also that calcium
formate has a very low solubility in water.5.1 Used at dosages of 2 to 3 per cent by
mass of cement, calcium formate increases the strength of concrete up to about 24
hours, the effect being greater with low C3A cements.5.3
Massazza and Testolin5.13 found that, with calcium formate, concrete could achieve
at hours the strength which would be reached only at 9 hours without the
admixture, as shown in the example of Fig. 5.2. It is useful to note that calcium
formate does not cause a retrogression of strength. On the other hand, the possibility
of side effects of this accelerator has not been eliminated.5.12,5.33

Fig. 5.2. Influence of calcium formate at various contents (by mass of cement) on
the development of strength of concrete with a cement content of 420 kg/m3 (710
lb/yd3) and a water/cement ratio of 0.35 (cited in ref. 5.13)

260
 Triethanolamine is a possible accelerator but it is very sensitive to dosage variation
and to the composition of cement.5.34 For this reason, triethanolamine is not used
except to offset the retarding action of some water-reducing admixtures.
The precise mode of action of accelerators is still unknown. Moreover, the effect
of accelerators on early strength of concrete very much depends on the particular
accelerator used, as well as on the cement used, even for cements of the same nominal
type. The full actual composition of the admixtures is usually not disclosed for
commercial reasons so that it is necessary to ascertain the performance of any
potential cement–admixture combination.
The extent of the problem was demonstrated by Rear and Chin5.20 who tested
concretes of the same mix proportions (water/cement ratio of 0.54) made with five
Type I Portland cements and three admixtures used at three dosages: No. 1 calcium-
nitrite-based; No. 2 calcium-nitrate-based; and No. 3 sodium-thiocyanate-based. The
range of compound composition of the cements (in per cent) was as follows:

The fineness of cement ranged from 327 to 429 m2/kg measured by the Blaine method.
From the resulting compressive strengths determined at 20 °C (72 °F), shown
in Table 5.1, it can be seen that there is a very wide variation in the performance of
each of the admixtures when used with the different cements, as well as between the
three admixtures. In all cases, the strength is expressed as a percentage of strength of
the accelerator-free concrete.

Table 5.1. Effect of Accelerators on the Strength of Concretes made with

Different Cements5.20

261
 The ASTM Specification C 494-10 includes a requirement that, when a Type C
admixture is used, the initial set, measured by the penetration resistance test
prescribed in ASTM C 403-10 be at least 1 hour earlier, but no more than hours
earlier, than that of a control mix. The compressive strength at 3 days is to be 125 per
cent of the strength of the control concrete. The strength beyond the age of 28 days is
allowed to be lower than the strength of the control concrete, but retrogression of
strength is not permitted. BS EN 934-2 : 2009 prescribes initial setting time, strength
and air content. That standard prescribes also the requirements for other types of
admixtures.
The preceding discussion indicates that no single accelerator is widely accepted. It
is useful to note, at the same time, that the demand for accelerators has decreased,
especially in the manufacture of precast concrete, as there exist other means of
achieving a high early strength, such as the use of very low water/cement ratios in
conjunction with superplasticizers. However, the use of accelerators at low placing
temperatures continues.
Retarding admixtures
A delay in the setting of the cement paste can be achieved by the addition to the mix
of a retarding admixture (ASTM Type B), for brevity, referred to as a retarder.
Retarders generally slow down also the hardening of the paste although some salts
may speed up the setting but inhibit the development of strength. Retarders do not
alter the composition or identity of products of hydration.5.45
Retarders are useful in concreting in hot weather, when the normal setting time is
shortened by the higher temperature, and in preventing the formation of cold joints. In
general, they prolong the time during which concrete can be transported, placed, and
compacted. The delay in hardening caused by the retarders can be exploited to obtain
an architectural finish of exposed aggregate: the retarder is applied to the interior

262
surface of the formwork so that the hardening of the adjacent cement is delayed. This
cement can be brushed off after the formwork has been struck so that an exposed
aggregate surface is obtained.
The use of retarders can sometimes affect structural design; for example,
continuous massive pours can be used with controlled retardation of the various parts
of the pour, instead of segmental construction (see p. 398).
Retarding action is exhibited by sugar, carbohydrate derivatives, soluble zinc salts,
soluble borates and some other salts;5.51 methanol is also a possible retarder.5.12 In
practice, retarders which are also water-reducing (ASTM Type B) are more
commonly used; these are described in the next section.
The mechanism of the action of retarders has not been established with certainty. It
is likely that they modify the crystal growth or morphology,5.37 becoming adsorbed on
the rapidly formed membrane of hydrated cement and slowing down the growth of
calcium hydroxide nuclei.5.11 These actions result in a more efficient barrier to further
hydration than is the case without an admixture. The admixtures are finally removed
from solution by being incorporated into the hydrated material but this does not
necessarily mean the formation of different complex hydrates.5.36 This is also the case
with water-reducing and retarding admixtures, that is ASTM Class D: Khalil and
Ward5.43 showed that the linear relation between the heat of hydration and the mass of
non-evaporable water is unaffected by the addition of a lignosulfonate-based
admixture (see Fig. 5.3).

263
Fig. 5.3. Relation between the non-evaporable water content of cement and heat
of hydration with and without a retarder5.43
Great care is necessary in using retarders because, in incorrect quantities, they can
totally inhibit the setting and hardening of concrete. Cases are known of seemingly
inexplicable results of strength tests when sugar bags have been used for the shipment
of aggregate samples to the laboratory or when molasses bags have been used to
transport freshly mixed concrete. The effects of sugar depend greatly on the quantity
used, and conflicting results were reported in the past.5.6 It seems that, used in a
carefully controlled manner, a small quantity of sugar (about 0.05 per cent of the mass
of cement) will act as an acceptable retarder: the delay in setting of concrete is about 4
hours.5.55 The retarding action of sugar is probably by the prevention of the formation
of C-S-H.5.50 However, the exact effects of sugar depend greatly on the chemical
composition of cement. For this reason, the performance of sugar, and indeed of any
retarder, should be determined by trial mixes with the actual cement which is to be
used in construction.
A large quantity of sugar, say 0.2 to 1 per cent of the mass of cement, will virtually
prevent the setting of cement. Such quantities of sugar can therefore be used as an
inexpensive ‘kill’, for instance when a mixer or an agitator has broken down and

264
cannot be discharged. In the construction of the tunnel between England and France in
the early 1990s, molasses was used to prevent setting of residual concrete as washing
out underground was not possible.
When sugar is used as a controlled set retarder, the early strength of concrete is
severely reduced5.26 but, beyond about 7 days, there is an increase in strength of several
per cent compared with a non-retarded mix.5.55 This is probably due to the fact that
delayed setting produces a denser hydrated cement gel (cf. p. 361).
It is interesting to note that the effectiveness of an admixture depends on the time
when it is added to the mix: a delay of even 2 minutes after water has come into
contact with the cement increases the retardation; sometimes, such a delay can be
achieved by a suitable sequence of feeding the mixer. The increased retardation
occurs especially with cements which have a high C3A content because, once some
C3A has reacted with gypsum, it does not adsorb the admixture so that more of it is
left to retard the hydration of the calcium silicates, which occurs through adsorption
onto the calcium hydroxide nuclei.5.36
As retarders are frequently used in hot weather it is important to note that the
retarding effect is smaller at higher temperatures (see Fig. 5.4) and some retarders
cease to be effective at extremely high ambient temperatures, about 60 °C
(140 °F).5.13 Fattuhi’s data5.10 on the effectiveness of various water-reducing and set-
retarding admixtures, in terms of the retardation of initial setting of concrete, are
given in Table 5.2; the effect of high temperature on the final setting time is much
smaller.

Fig. 5.4. Influence of temperature on initial setting time of concretes with various
contents of retarder (by mass of cement) (cited in ref. 5.13)

Table 5.2. Influence of Air Temperature on the Retardation of the Initial Setting
Time of Concrete* by Water-reducing and Set-retarding
Admixtures5.10 (Copyright ASTM-reproduced with permission)

265
 *
Measured by penetration resistance according to ASTM C 403-08.
Retarders tend to increase the plastic shrinkage because the duration of the plastic
stage is extended, but drying shrinkage is not affected.5.38
ASTM C 494-10 requires Type B admixtures to retard the initial set by at least 1
hour but not more than hours, as compared with a control mix. The compressive
strength from the age of 3 days onwards is allowed to be 10 per cent less than the
strength of the control concrete. The requirements of BS 5075-1 : 1982 are broadly
similar. A specification for various types of admixtures is given in BS EN 934-2 :
2009 and ASTM C 494-10.
Water-reducing admixtures
According to ASTM C 494-10, admixtures which are only water-reducing are called
Type A, but if the water-reducing properties are associated with retardation, the
admixture is classified as Type D. There exist also water-reducing and accelerating
admixtures (Type E) but these are of little interest. However, if the water-reducing
admixture exhibits, as a side effect, set retardation, this can be combated by an
integral incorporation of an accelerator in the mix. The most common accelerator is
triethanolamine (see p. 251).
As their name implies, the function of water-reducing admixtures is to reduce the
water content of the mix, usually by 5 or 10 per cent, sometimes (in concretes of very
high workability) up to 15 per cent. Thus, the purpose of using a water-reducing
admixture in a concrete mix is to allow a reduction in the water/cement ratio while
retaining the desired workability or, alternatively, to improve its workability at a
given water/cement ratio. Whereas aggregate which is manifestly badly graded should
not be used, water-reducing admixtures improve the properties of fresh concrete made
with poorly graded aggregate, e.g. a harsh mix (see pp. 165 and 747). Concrete
containing a water-reducing admixture generally exhibits low segregation and good
‘flowability’.
Water-reducing admixtures can also be used in pumped concrete or in concrete
placed by a tremie.
The two main groups of admixtures of Type D are: (a) lignosulfonic acids and
their salts, and (b) hydroxylated carboxylic acids and their salts. The modifications
and derivatives of these do not act as retarders, and may even behave as
accelerators5.28 (see Fig. 5.5): they are therefore of Type A or E (see p. 246).

266
 Fig. 5.5. Effect of various water-reducing admixtures on the setting time of
concrete.5.28 Numbers 1 and 2 are lignosulfonate-based; 3 and 4 are hydroxylated
carboxylic acid-based
The principal active components of the admixtures are surface-active
agents.5.27 These are substances which are concentrated at the interface between two
immiscible phases and which alter the physico-chemical forces acting at this interface.
The substances are adsorbed on the cement particles, giving them a negative charge
which leads to repulsion between the particles, that is to their deflocculation, and
results in stabilizing their dispersion; air bubbles are also repelled and cannot attach
themselves to the cement particles. Because flocculation traps some water, and also
because where cement particles touch one another, their touching surfaces are not
available for early hydration, water-reducing admixtures increase the surface area of
cement which can undergo initial hydration and also increase the amount of water
available for hydration.
In addition, the electrostatic charge causes the development around each particle of
a sheath of oriented water molecules which prevent a close approach of the particles
to one another. The particles have, therefore, a greater mobility, and water freed from
the restraining influence of the flocculated system becomes available to lubricate the
mix so that the workability is increased.5.27 Some Type D admixtures are also adsorbed
on the products of hydration of cement.
As one effect of dispersion of cement particles, already mentioned, is to expose a
greater surface area of cement to hydration, which progresses therefore at a higher
rate in the early stages, there is an increase in the strength of concrete, compared with
a mix of the same water/cement ratio but without the admixture. A more uniform
distribution of the dispersed cement throughout the concrete may also contribute to a
higher strength5.27 because the process of hydration is improved. The increase in
strength is particularly noticeable in very young concretes5.29 but under certain
conditions persists for a long time.
Although water-reducing admixtures affect the rate of hydration of cement, the
nature of the products of hydration is unchanged5.33 and so is the structure of the
hydrated cement paste. Consequently, the use of water-reducing admixtures does not
affect the resistance of concrete to freezing and thawing.5.2 This statement is valid on

267
condition that the water/cement ratio is not increased in conjunction with the use of
the admixture. More generally, in assessing the benefits of the use of water-reducing
admixtures, it is vital to use a proper base for any comparison and not simply to rely
on commercial assertions. It should be noted that, even though some water-reducing
admixtures may result in set retardation, they do not always reduce the rate of loss of
workability with time.5.29 A further aspect to be considered is the danger of segregation
of the concrete and of bleeding.
The effectiveness of water-reducing admixtures with respect to strength varies
considerably with the composition of cement, being greatest when used with cements
of low alkali or low C3A content. An example of the improvement in workability of
mix with a given water content and a given dosage of lignosulfonate admixture, as a
function of the C3A content of the Portland cement used, cited by Massazza and
Testolin,5.13 is shown in Fig. 5.6.

Fig. 5.6. Influence of the content of C3A in the cement (at a constant ratio of C3S
to C2S) on the increase in the flow of mortar (over the flow of an admixture-free
mortar) at a 0.2 per cent dosage of a lignosulfonate admixture (cited in ref. 5.13)
Generally, the dosage per 100 kg of cement is lower in mixes with a high cement
content. Some water-reducing admixtures are more effective when used in mixes
containing pozzolanas than in Portland-cement-only mixes.
Whereas an increased dosage of water-reducing admixture increases the
workability5.2 (see Fig. 5.7), there would be an associated considerable retardation,
which is likely to be unacceptable. Long-term strength, however, is unaffected.5.28

268
 Fig. 5.7. Influence of dosage of retarders on slump (based on ref. 5.2)
With many water-reducing admixtures, a slight delay in the introduction of the
admixtures into the mix (even as low as 20 seconds from the time of contact between
cement and water) enhances the performance of the admixture.
The dispersing action of a water-reducing admixture has also some effect on the
dispersion of air in water5.1 so that the admixture, especially lignosulfonate-based, may
have some air-entraining effect. As this results in a reduction in the strength of
concrete (see p. 561), the effect is undesirable; on the other hand, the entrained air
improves the workability. Air entrainment can be counteracted by the inclusion of a
small amount of a detraining agent in the water-reducing admixture; the usual agent is
tributylphosphate.5.2
Lignosulfonate-based water-reducing admixtures increase shrinkage, but other
water-reducing admixtures have been shown not to affect shrinkage.5.13
With some cements, the influence of water-reducing admixtures is very small but,
in general terms, admixtures are effective with all types of Portland cement and also
with high-alumina cement. The actual effectiveness of any water-reducing admixture
depends on the cement content, water content, type of aggregate used, presence of air-
entraining agents or pozzolanas, as well as on temperature. It is, therefore, apparent
that trial mixes, containing the actual materials to be used on the job, are essential in
order to determine the type and quantity of admixture to achieve optimum properties:
reliance on the data given by the manufacturers of admixtures is insufficient.
Superplasticizers
Superplasticizers are admixtures which are water reducing but significantly and
distinctly more so than the water-reducing admixtures considered in the preceding

269
section. Superplasticizers are also usually highly distinctive in their nature, and they
make possible the production of concrete which, in its fresh or hardened state, is
substantially different from concrete made using water-reducing admixtures of Types
A, D, or E because a very low w/c or a high workability can be obtained.
For these reasons, superplasticizers are classified separately by ASTM C 494-10,
and they are discussed separately in this book. ASTM C 494-10 refers to
superplasticizers as “water-reducing, high range admixtures” but this name seems to
be too long and too complex. On the other hand, it has to be admitted that the name
‘superplasticisers’ smacks of ‘super’-commercialism, but it has become widely
accepted and has, at least, the merit of brevity. In this book, therefore, the term
superplasticizers will be used.
In the ASTM terminology, superplasticizers are referred to as Type F admixtures;
when the superplasticizers are also retarding they are called Type G admixtures.
Nature of superplasticizers
There exist four main categories of superplasticizers: sulfonated melamine-
formaldehyde condensates; sulfonated naphthalene-formaldehyde condensates;
modified lignosulfonates; and others such as sulfonic-acid esters and carbohydrate
esters.
The first two are the most common ones. For brevity, they will be referred to as
melamine-based superplasticizers and naphthalene-based superplasticizers,
respectively.
Superplasticizers are water-soluble organic polymers which have to be synthesized,
using a complex polymerization process, to produce long molecules of high molecular
mass, and they are therefore relatively expensive. On the other hand, because they are
manufactured for a specific purpose, their characteristics can be optimized in terms of
length of molecules with minimum cross-linking. They also have a low content of
impurities so that, even at high dosages, they do not exhibit unduly harmful side
effects.
A larger molecular mass, within limits, improves the efficiency of superplasticizers.
Their chemical nature also has an effect, but no generalizations about the overall
superiority of either naphthalene- or melamine-based superplasticizers is possible,
probably because more than one property of a superplasticizer affects its performance
and also because the chemical properties of the cement used play a role as well.5.21
The majority of superplasticizers are in the form of sodium salts but calcium salts
are also produced; the latter, however, have a lower solubility. A consequence of the
use of sodium salts is the introduction of additional alkalis into the concrete which
may be relevant to the reactions of hydration of the cement and to a potential alkali–
silica reaction. For this reason, the soda content of the admixtures should be known;
in some countries, e.g. Germany, the content is limited to 0.02 per cent of soda by
mass of cement.5.22
A modification of the naphthalene-based superplasticizer by the inclusion of a
copolymer with a functional sulfonic group and carboxyl group has been
developed.5.35 This maintains the electrostatic charge on the cement particles and
prevents flocculation by adsorption on the surface of cement particles. The copolymer
is more active at higher temperatures, which is particularly beneficial in concreting in
hot weather when high workability can be retained for up to one hour after mixing.5.35

270
 When adequate information about the detailed nature of a superplasticizer is not
provided, much can be learnt from specialized chemical tests.5.15
Physical tests make it possible readily to distinguish superplasticizers from water-
reducing admixtures.5.16
Effects of superplasticizers
The main action of the long molecules is to wrap themselves around the cement
particles and give them a highly negative charge so that they repel each other or act by
steric repulsion. This results in deflocculation and dispersion of cement particles. The
resulting improvement in workability can be exploited in two ways: by producing
concrete with a very high workability or concrete with a very high strength.
At a given water/cement ratio and water content in the mix, the dispersing action
of superplasticizers increases the workability of concrete, typically by raising the
slump from 75 mm (3 in.) to 200 mm (8 in.), the mix remaining cohesive (see Fig.
5.8).5.42 Even a higher slump can be achieved in self-compacting concrete. The
resulting concrete can be placed with little or no compaction and is not subject to
excessive bleeding or segregation. Such concrete is termed flowing concrete and is
useful for placing in very heavily reinforced sections, in inaccessible areas, in floor
slabs, and also where very rapid placing is desired. Properly compacted flowing
concrete is believed to develop normal bond with reinforcement.5.52 It should be
remembered, when designing formwork, that flowing concrete can exert full
hydrostatic pressure.

Fig. 5.8. Relation between flow table spread and water content of concrete with
and without superplasticizer5.42

271
 The second use of superplasticizers is in the production of concrete of normal
workability but with an extremely high strength owing to a very substantial reduction
in the water/cement ratio. Water/cement ratios down to 0.2 have been used with 28-
day cylinder strengths of about 150 MPa (22 000 psi). Generally speaking,
superplasticizers can reduce the water content for a given workability by 25 to 35 per
cent (compared with less than half that value in the case of conventional water-
reducing admixtures), and increase the 24-hour strength by 50 to 75 per cent;5.39 an
even greater increase occurs at somewhat earlier ages. Practical mixes with a cube
strength of 30 MPa (4300 psi) at 7 hours have been obtained5.39 (see Fig. 5.9). With
steam-curing or high-pressure steam-curing, even higher early strengths are possible.

Fig. 5.9. The influence of the addition of superplasticizer on the early strength
(measured on cubes) of concrete with a cement content of 370 kg/m3 (630 lb/yd3)
and cast at room temperature. Type III cement; all concretes of the same
workability5.46
Performance requirements for superplasticizers to produce flowing concrete and to
produce very high strength concrete are given, respectively, in ASTM C 1017-07 and
ASTM C 494-10, and for both types of concrete in BS EN 934-2 : 2009. It is worth
noting that the Standard requirements for improvement both in workability and in
strength are greatly exceeded by the available commercial superplasticizers.
Superplasticizers do not alter fundamentally the structure of hydrated cement paste,
the main effect being a better distribution of cement particles and, consequently, their
better hydration. This would explain why, in some cases, the use of superplasticizers
was found to increase the strength of concrete at a constant water/cement ratio. Values

272
of a 10 per cent increase at 24 hours and a 20 per cent increase at 28 days have been
quoted, but this behaviour has not been universally confirmed.5.13
What is important is that no retrogression of strength at long ages has ever been
reported.
While the mechanism of the action of superplasticizers has not been fully
explained, it is known that they interact with C3A whose hydration is retarded.5.13 A
physical consequence is the formation of ettringite crystals which are small and nearly
cubic in shape rather than needle-like. The cubic shape improves the mobility of the
cement paste,5.21 but is unlikely to be the main mechanism of action of
superplasticizers because they also improve the workability of partially hydrated
cement in which the ettringite crystals are already formed. The ultimate fate of
superplasticizers is not completely known.5.49
Some superplasticizers do not produce appreciable set retardation, but there exist
also set-retarding superplasticizers, classified by ASTM C 494-10 as Type G. In cases
of naphthalene-based superplasticizers where retardation was observed, Aïtcin et
al.5.5 showed that this applies principally to the cement particles in the size range of 4
to 30 μm. Particles smaller than 4 μm are not affected as they are rich in SO3 and in
the alkalis; large particles undergo little initial hydration regardless of the presence or
absence of a superplasticizer.5.5
Because superplasticizers do not significantly affect the surface tension of water,
they do not entrain large amounts of air and can therefore be used at high dosages.
Dosage of superplasticizers
For increasing the workability of the mix, the normal dosage of superplasticizers is
between 1 and 3 litres per cubic metre of concrete, the liquid superplasticizer
containing about 40 per cent of active material. When superplasticizers are used to
reduce the water content of the mix, their dosage is much higher: 5 to 20 litres per
cubic metre of concrete. In the calculation of the water/cement ratio and of mix
proportions in general, the volume of the liquid containing superplasticizer must be
taken into account.
It is worth noting that the concentration of solids in commercial superplasticizers
varies so that any comparison of performance should be made on the basis of the
amount of solids, and not on the total mass. For practical purposes, comparison should
be made on the basis of the price for a given effect.
The effectiveness of a given dosage of a superplasticizer depends on the
water/cement ratio of the mix. Specifically, at a given dosage of the superplasticizer,
the percentage water reduction which maintains a constant workability is much higher
at low water/cement ratios than at high water/cement ratios; for example, at a
water/cement ratio of 0.40, the reduction was observed to be 23 per cent, and only 11
per cent at a water/cement ratio of 0.55.5.13
When superplasticizers are used in very low dosages to produce high-workability
normal-strength concrete, there are few problems in selecting an admixture–cement
combination. At high dosages, the situation is significantly different in that the
superplasticizer has to be compatible with the actual cement used, and it is not enough
for the superplasticizer and the cement separately to conform to their respective
standards. The problem of compatibility is discussed on p. 680.
Loss of workability

273
It is logical to assume that the first dosage of the superplasticizer must be applied
soon after the cement and water have come into contact with one another. Otherwise,
the initial reactions of hydration would make it impossible for the superplasticizer to
effect adequate deflocculation of the cement particles. Data at variance with the
preceding statement have been reported but not explained.5.1
The theoretical optimum time for adding a superplasticizer is what would be
approximately the beginning of the dormant period without the superplasticizer. In
fact, addition at that time was found to result in the highest initial workability and in
the lowest rate of loss of workability.5.30 This particular time depends on the properties
of cement and would have to be established by experiment. In actual construction, it is
the practicality of adding the superplasticizer that governs.
The effectiveness of superplasticizers in preventing re-agglomeration of cement
particles lasts only as long as sufficient superplasticizer molecules are available to
cover the exposed surface of cement particles. As some of the superplasticizer
molecules become entrapped in the products of hydration of cement or react with the
C3A, the supply of superplasticizer becomes inadequate and the workability of the mix
is rapidly lost. It is likely that, with prolonged mixing or agitation, some of the
products of initial hydration of the cement shear off the surface of the cement particles.
This enables the hydration of the hitherto unexposed cement to take place. Both the
presence of the detached products of hydration and the additional hydration have the
effect of reducing the workability of the mix.
An example5.31 of the loss of workability of concrete made with a naphthalene-
based superplasticizer is shown in Fig. 5.10. For comparison, the loss of workability
of an admixture-free mix with the same initial slump is shown in the same figure. It
can be seen that the loss occurs much faster with a superplasticizer but, of course, the
superplasticized concrete has a lower water/cement ratio and consequently a higher
strength.

274
 Fig. 5.10. Loss of slump with time of concretes: (A) water/cement ratio of 0.58
and no admixture; (B) water/cement ratio of 0.47 and superplasticizer (based on
ref. 5.31)
Because the effectiveness of superplasticizers is limited in duration, it may be
advantageous to add the superplasticizer to the mix in two, or even three, operations.
Such repeated addition, or re-dosage, is possible if an agitator truck is used to deliver
the concrete to site. If the workability is to be restored by the re-dosage some time
after the original mixing, the amount of superplasticizer has to be adequate to act both
on the cement particles and on the products of hydration. Therefore, a high re-dosage
of superplasticizer is necessary; a small re-dosage is ineffective.5.23
Whereas repeated addition of the superplasticizer to the mix is beneficial from the
standpoint of workability, it may increase bleeding and segregation. Other possible
side effects are set retardation and a change (up or down) in the amount of entrained
air.5.4 Also, the workability restored by the second dosage may decrease at a fast rate
so that the re-dosage should preferably be applied immediately prior to placing and
compaction of the concrete.
An example of an effect of the application of a re-dosage of a naphthalene-based
superplasticizer on workability is shown in Fig. 5.11 for a concrete with a
water/cement ratio of 0.50; the initial dosage and each of the subsequent three re-
dosages were the same, namely 0.4 per cent of solids by mass of cement.

275
Fig. 5.11. Influence of repeated re-dosage of naphthalene-based superplasticizer
on slump (based on ref. 5.1)
The quantity of superplasticizer which needs to be added to restore the workability
increases with temperature in the range of 30 to 60 °C (86 to 140 °F), and is much
higher at a water/cement ratio of about 0.4 than at higher water/cement ratios. Even
though the original workability is restored by a second or even a third dosage of a
superplasticizer, the subsequent loss of workability becomes more rapid. However,
the rate of the loss is not increased at higher temperatures.5.18
Nowadays, there exist superplasticizers with a long period of effectiveness so that
re-dosing immediately prior to placing of concrete can be avoided. The use of such
superplasticizers offers a better control of the mix proportions and is, therefore,
preferable.5.52
Superplasticizer–cement compatibility
If a large dosage of the superplasticizer is used in order to achieve a very low
water/cement ratio or if re-dosage of the superplasticizer is not possible, it is
important to establish a compatible superplasticizer–cement combination. When the
two materials are well-matched, a large single dosage can lead to the retention of high
workability for a sufficiently long period: 60 to 90 minutes can be achieved and,
occasionally, even 2 hours.
While assessing compatibility, the required dosage of the superplasticizer should
be established. The usual approach is to determine the percentage water reduction
which will result in the same workability as an admixture-free mix, using the flow-
table method of ASTM C 230-08 or BS 1881-105 : 1984 (withdrawn). Alternatively,
a mini-slump test developed by Kantro5.54 can be used. Aïtcin et al.5.21 favour the use of

276
a Marsh cone for the determination of the time required for a specified volume of
grout containing the given cement and superplasticizer to flow through an orifice.
Generally, this time, known as Marsh flow-time, decreases with an increase in the
superplasticizer dosage up to a value beyond which there is little improvement. This is
the optimum dosage. Apart from reasons of economy, an excessive dosage of
superplasticizer is undesirable as it leads to segregation. Also, there should be very
little difference in workability (as measured by Marsh flow-time) at 5 and at 60
minutes after mixing. Full discussion of this topic is given on p. 680.
The laboratory determination of the superplasticizer dosage should be followed by
a full-scale test but is nevertheless very valuable in rapidly verifying the suitability of
a given superplasticizer with a given cement. Several properties of cement are
relevant. For example, the finer the cement the higher the dosage of a superplasticizer
required to obtain a given workability.5.17 The chemical properties of cement, such as a
high C3A content (which reduces the effectiveness of a given dosage of the
superplasticizer) and the nature of calcium sulfate used as a retarder, also affect the
performance of superplasticizers.5.21
From the preceding discussion, it can be seen that a single value of dosage,
sometimes recommended by the superplasticizer manufacturer, is of little value.
In searching for a suitable combination of cement and superplasticizer, it is
sometimes easier to vary the superplasticizer whereas, at other times, there is a
selection of cements available; what must not be assumed is that any indiscriminate
combination of the two materials will be satisfactory. Reliable means of establishing
compatibility of a Portland cement and a superplasticizer are available.5.17
Use of superplasticizers
The availability of superplasticizers has revolutionized the use of concrete in a
number of ways, making it possible to place it, and to do so easily, where it was not
possible to do so before. Superplasticizers make it also possible to produce concrete
with significantly superior strength and other properties, now termed high
performance concrete (see Chapter 13).
Superplasticizers do not significantly affect the setting time of concrete except that,
when used with cements having a very low C3A content, there may be high retardation.
They can be successfully used in concrete containing fly ash5.47 and are particularly
valuable when silica fume is present in the mix because that material increases the
water demand of the mix.5.32 However, if re-dosage is necessary, the quantity of the
superplasticizer required is larger than when the concrete contains no silica fume.5.19
Superplasticizers do not influence shrinkage, creep, modulus of elasticity5.41 or
resistance to freezing and thawing.5.40 They have no effect per se on the durability of
concrete.5.14 Specifically, durability on exposure to sulfates is unaffected.5.41 The use of
superplasticizers with an air-entraining admixture requires caution as sometimes the
actual amount of entrained air is modified by the superplasticizer. The influence of
superplasticizers on air entrainment and on the resulting resistance of concrete to
freezing and thawing is considered on p. 554.
Special admixtures
In addition to the admixtures so far considered in the present chapter, there exist also
admixtures for other purposes, such as air detrainment, anti-bacterial action, and
waterproofing, but these are not sufficiently standardized to make reliable

277
generalizations possible. Moreover, some of the names under which certain
admixtures are sold give an exaggerated impression of their performance.
This is not to say that these admixtures are not beneficial: under many
circumstances, they serve a very useful purpose, but their performance needs to be
carefully established prior to use.
Waterproofing admixtures
Concrete absorbs water because surface tension in capillary pores in the hydrated
cement paste ‘pulls in’ water by capillary suction. Waterproofing admixtures aim at
preventing this penetration of water into concrete. Their performance is very much
dependent on whether the applied water pressure is low, as in the case of rain (other
than driven by wind) or capillary rise, or whether a hydrostatic pressure is applied, as
in the case of water-retaining structures or structures such as basements in
waterlogged ground. The term ‘waterproofing’ is therefore of dubious validity.
Waterproofing admixtures may act in several ways but their effect is mainly to
make concrete hydrophobic. By this is meant an increase in the contact angle between
the walls of the capillary pores and water, so that water is ‘pushed out’ of the pores.
One action of waterproofing admixtures is through reaction with the calcium
hydroxide in hydrated cement paste; examples of products used are stearic acid and
some vegetable and animal fats. The effect is to make the concrete hydrophobic.
Another action of waterproofing admixtures is through coalescence on contact
with the hydrated cement paste which, because of its alkalinity, breaks down the
‘waterproofing’ emulsion; an example is an emulsion of very finely divided wax. The
effect here, too, is to make the concrete hydrophobic.
The third type of waterproofing admixture is in the form of very fine material
containing calcium stearate or some hydrocarbon resins or coal tar pitches which
produce hydrophobic surfaces.5.2
While imparting hydrophobic properties to concrete is valuable, in practice,
complete coating of all surfaces of capillary pores is difficult to attain, with the
consequence that full waterproofing is unlikely to be achieved.5.3
Some waterproofing admixtures, in addition to their hydrophobic action, also
effect pore blocking through a coalescent component. Unfortunately, little
information is available to make it possible to explain and classify the actions
involved so that reliance has to be based on the manufacturers’ data coupled with
experimental evidence on the performance of any particular waterproofing admixture.
It should be stressed that the experience should be over a sufficiently long period to
demonstrate the stability of the waterproofing admixture.
A side effect of some waterproofing admixtures is to improve the workability of
the mix owing to the presence of finely divided wax or bituminous emulsions, which
entrain some air. They also improve cohesion of the concrete but may result in a
‘sticky’ mix.5.3
Because of the nature of the waterproofing admixtures, they are not effective in
resisting attack by aggressive gases.5.2
A final point to be made about waterproofing admixtures is that, because their
exact composition is often unknown, it is vital to ascertain that they contain no
chlorides if the concrete is likely to be used in a situation which is sensitive to
chloride-induced corrosion.

278
 Waterproofing admixtures should be distinguished from water repellents, based on
silicone resins, which are applied to the concrete surface. Waterproof membranes are
emulsion-based bitumen coatings, possibly with rubber latex, which produce a tough
film with some degree of elasticity. Consideration of these materials is outside the
scope of this book.
Anti-bacterial and similar admixtures
Some organisms such as bacteria, fungi or insects can adversely affect concrete. The
possible mechanisms5.3 are: releasing corrosive chemicals through metabolic action,
and creation of an environment which promotes corrosion of steel. Staining of the
surface can also result.
The usual agent in bacterial attack is an organic or mineral acid which reacts with
hydrated cement paste. Initially, the alkaline pore water in hydrated cement paste
neutralizes the acid but continuing action of bacteria results in deeper attack.
Because the rough surface texture of concrete shelters the bacteria, surface
cleaning is ineffective, and it is necessary to incorporate in the mix some special
admixtures which are toxic to the attacking organisms: these may be anti-bacterial,
fungicidal or insecticidal.
Fuller details of bacterial attack are given by Ramachandran.5.3 Useful information
about anti-bacterial admixtures is given in ACI 212.3R-91,5.4 which lists some
effective admixtures. It can be added that copper sulfate and pentachlorophenol have
been found to control the growth of algae or lichen on hardened concrete but their
effectiveness is lost with time.5.48 Clearly, admixtures which may prove toxic should
not be used.
Conversely, some bacteria introduced into the concrete mix may heal cracks by
way of precipitating calcite. These bacteria are spore-forming and alkali
resistant.5.56 Laboratory experiments demonstrated successful healing of cracks when
water enters the cracks. However, a substrate of organic carbon is necessary, but such
carbon in the mix may be deleterious. Thus further studies are necessary. Also, there
is a question of cost of incorporation of the bacteria in the mix.
Remarks about the use of admixtures
Admixtures whose performance is known from experience at normal ambient
temperatures may behave differently at very high or very low temperatures.
Some admixtures do not tolerate exposure to freezing temperatures while stored
and become unusable; most of the others require thawing and remixing. Very few are
unaffected by freezing temperatures.
Admixtures, whose performance when used separately is known, may not be
compatible when used together; for this reason, it is essential to use trial mixes for any
combination of admixtures.
Even if two admixtures are compatible when introduced into the mix, they may
interact adversely if they come into contact with one another prior to being introduced
into the mixer. This is, for example, the case with the combination of a water-
reducing admixture of the lignosulfonate type and an air-entraining admixture of a
vinsol resin-based type.5.29 In consequence, it is a wise precaution to discharge the
various admixtures into the mixer separately and at different locations, and possibly
also at different times. Details of admixture batching systems are given in ACI
212.3R-1991.5.4

279
 When being discharged into the mixer, admixtures have to be not only accurately
metered, but it is also important that they be discharged in the correct part of the
mixing cycle and at the correct rate. Changes in concrete mixing procedure can affect
the performance of admixtures.
It is important to know whether any admixture to be used contains chlorides
because, generally, there is specified a limit on the total chloride ion content in the
concrete mix so that all sources of chlorides have to be taken into account
(see Chapter 11). Even the so-called ‘chloride-free’ admixtures may contain small
amounts of chloride ions originating from the water used in the manufacture of the
admixture. When there is high sensitivity to the chloride content of the concrete, for
instance for use in prestressed concrete, the exact chloride content of the admixture to
be used should be ascertained.5.4
References
5.1. V. DODSON, Concrete Admixtures, 211 pp. (New York, Van Nostrand
Reinhold, 1990).
5.2. M. R. RIXOM and N. P. MALIVAGANAM, Chemical Admixtures for Concrete, 2nd
Edn, 306 pp. (London/New York, E. & F. N. Spon, 1986).
5.3. V. S. RAMACHANDRAN, Ed., Concrete Admixtures Handbook: Properties,
Science and Technology, 626 pp. (New Jersey, Noyes Publications, 1984).
5.4. ACI 212.3R-91, Chemical admixtures for concrete, in ACI Manual of
Concrete Practice, Part 1: Materials and General Properties of Concrete,
31 pp. (Detroit, Michigan, 1994).
5.5. P.-C. AÏTCIN, S. L. SARKAR, M. REGOURD and D. VOLANT, Retardation effect of
superplasticizer on different cement fractions, Cement and Concrete
Research, 17, No. 6, pp. 995–9 (1987).
5.6. F. M. LEA, The Chemistry of Cement and Concrete (London, Arnold, 1970).
5.7. S. GEBLER, Evaluation of calcium formate and sodium formate as accelerating
admixtures for portland cement concrete, ACI Journal, 80, No. 5, pp. 439–
44 (1983).
5.8. K. SAKAI, H. WATANABE, H. NOMACI and K. HAMABE, Preventing freezing of
fresh concrete, Concrete International, 13, No. 3, pp. 26–30 (1991).
5.9. C. J. KORHONEN and E. R. CORTEZ, Antifreeze admixtures for cold weather
concreting, Concrete International, 13, No. 3, pp. 38–41 (1991).
5.10. N. J. FATTUHI, Influence of air temperature on the setting of concrete
containing set retarding admixtures, Cement, Concrete and Aggregates, 7,
No. 1, pp. 15–18 (Summer 1985).
5.11. P. F. G. BANFILL, The relationship between the sorption of organic
compounds on cement and the retardation of hydration, Cement and
Concrete Research, 16, No. 3, pp. 399–410 (1986).
5.12. V. S. RAMACHANDRAN and J. J. BEAUDOIN, Use of methanol as an admixture, Il
Cemento, 84, No. 2, pp. 165–72 (1987).
5.13. F. MASSAZA and M. TESTOLIN, Latest developments in the use of admixtures
for cement and concrete, Il Cemento, 77, No. 2, pp. 73–146 (1980).
5.14. V. M. MALHOTRA, Superplasticizers: a global review with emphasis on
durability and innovative concretes, in Superplasticizers and Other

280
 Chemical Admixtures in Concrete, Proc. Third International Conference,
Ottawa, Ed. V. M. Malhotra, ACI SP-119, pp. 1–17 (Detroit, Michigan,
1989).
5.15. E. ISTA and A. VERHASSELT, Chemical characterization of plasticizers and
superplasticizers, in Superplasticizers and Other Chemical Admixtures in
Concrete, Proc. Third International Conference, Ottawa, Ed. V. M. Malhotra,
ACI SP-119, pp. 99–116 (Detroit, Michigan, 1989).
5.16. A. VERHASSELT and J. PAIRON, Rapid methods of distinguishing plasticizer from
superplasticizer and assessing superplasticizer dosage, in Superplasticizers
and Other Chemical Admixtures in Concrete, Proc. Third International
Conference, Ottawa, Ed. V. M. Malhotra, ACI SP-119, pp. 133–56 (Detroit,
Michigan, 1989).
5.17. E. HANNA, K. LUKE, D. PERRATON and P.-C. AÏTCIN, Rheological behavior of
portland cement in the presence of a superplasticizer, in Superplasticizers
and Other Chemical Admixtures in Concrete, Proc. Third International
Conference, Ottawa, Ed. V. M. Malhotra, ACI SP-119, pp. 171–88 (Detroit,
Michigan, 1989).
5.18. M. A. SAMARAI, V. RAMAKRISHNAN and V. M. MALHOTRA, Effect of retempering
with superplasticizer on properties of fresh and hardened concrete mixed at
higher ambient temperatures, in Superplasticizers and Other Chemical
Admixtures in Concrete, Proc. Third International Conference, Ottawa, Ed.
V. M. Malhotra, ACI SP-119, pp. 273–96 (Detroit, Michigan, 1989).
5.19. A. M. PAILLÈRE and J. SERRANO, Influence of dosage and addition method of
superplasticizers on the workability retention of high strength concrete with
and without silica fume (in French), in Admixtures for Concrete:
Improvement of Properties, Proc. ASTM Int. Symposium, Barcelona, Spain,
Ed. E. Vázquez, pp. 63–79 (London, Chapman and Hall, 1990).
5.20. K. REAR and D. CHIN, Non-chloride accelerating admixtures for early
compressive strength, Concrete International, 12, No. 10, pp. 55–8 (1990).
5.21. P.-C. AÏTCIN, C. JOLICOEUR and J. G. MACGREGOR, A look at certain
characteristics of superplasticizers and their use in the industry, Concrete
International, 16, No. 15, pp. 45–52 (1994).
5.22. T. A. BÜRGE and A. RUDD, Novel admixtures, in Cement Admixtures, Use and
Applications, 2nd Edn, Ed. P. C. Hewlett, for The Cement Admixtures
Association, pp. 144–9 (Harlow, Longman, 1988).
5.23. D. RAVINA and A. MOR, Effects of superplasticizers, Concrete International, 8,
No. 7, pp. 53–5 (July 1986).
5.24. J. J. SHIDELER, Calcium chloride in concrete, J. Amer. Concr. Inst., 48, pp.
537–59 (March 1952).
5.25. A. G. A. SAUL, Steam curing and its effect upon mix design, Proc. of a
Symposium on Mix Design and Quality Control of Concrete, pp. 132–42
(London, Cement and Concrete Assoc., 1954).
5.26. D. L. BLOEM, Preliminary tests of effect of sugar on strength of mortar, Nat.
Ready-mixed Concr. Assoc. Publ. (Washington DC, August 1959).

281
 5.27. M. E. PRIOR and A. B. ADAMS, Introduction to producers’ papers on water-
reducing admixtures and set-retarding admixtures for concrete, ASTM Sp.
Tech. Publ. No. 266, pp. 170–9 (1960).
5.28. C. A. VOLLICK, Effect of water-reducing admixtures and set-retarding
admixtures on the properties of plastic concrete, ASTM Sp. Tech. Publ. No.
266, pp. 180–200 (1960).
5.29. B. FOSTER, Summary: Symposium on effect of water-reducing admixtures
and set-retarding admixtures on properties of concrete, ASTM Sp. Tech. Publ.
No. 266, pp. 240–6 (1960).
5.30. G. CHIOCCHIO, T. MANGIALARDI and A. E. PAOLINI, Effects of addition time of
superplasticizers in workability of portland cement pastes with different
mineralogical composition, Il Cemento, 83, No. 2, pp. 69–79 (1986).
5.31. S. H. GEBLER, The effects of high-range water reducers on the properties of
freshly mixed and hardened flowing concrete, Research and Development
Bulletin RD081.01T, Portland Cement Association, 12 pp. (1982).
5.32. T. MANGIALARDI and A. E. PAOLINI, Workability of superplasticized
microsilica–Portland cement concretes, Cement and Concrete Research, 18,
No. 3, pp. 351–62 (1988).
5.33. P. C. HEWLETT, Ed., Cement Admixtures, Use and Applications, 2nd Edn, for
The Cement Admixtures Association, 166 pp. (Harlow, Longman, 1988).
5.34. J. M. DRANSFIELD and P. EGAN, Accelerators, in Cement Admixtures, Use and
Applications, 2nd Edn, Ed. P. C. Hewlett, for The Cement Admixtures
Association, pp. 102–29 (Harlow, Longman, 1988).
5.35. K. MITSUI et al., Properties of high-strength concrete with silica fume using
high-range water reducer of slump retaining type, in Superplasticizers and
Other Chemical Admixtures in Concrete, Ed. V. M. Malhotra, ACI SP-119,
pp. 79–97 (Detroit, Michigan, 1989).
5.36. J. F. YOUNG, A review of the mechanisms of set-retardation of cement pastes
containing organic admixtures, Cement and Concrete Research, 2, No. 4, pp.
415–33 (1972).
5.37. J. F. YOUNG, R. L. BERGER and F. V. LAWRENCE, Studies on the hydration of
tricalcium silicate pastes. III Influence of admixtures on hydration and
strength development, Cement and Concrete Research, 3, No. 6, pp. 689–
700 (1973).
5.38. C. F. SCHOLER, The influence of retarding admixtures on volume changes in
concrete, Joint Highway Res. Project Report JHRP-75-21, 30 pp. (Purdue
University, Oct. 1975).
5.39. P. C. HEWLETT and M. R. RIXOM, Current practice sheet No. 33 –
superplasticized concrete, Concrete, 10, No. 9, pp. 39–42 (London, 1976).
5.40. V. M. MALHOTRA, Superplasticizers in concrete, CANMET Report MRP/MSL
77-213, 20 pp. (Canada Centre for Mineral and Energy Technology, Ottawa,
Aug. 1977).
5.41. J. J. BROOKS, P. J. WAINWRIGHT and A. M. NEVILLE, Time-dependent properties
of concrete containing a superplasticizing admixture, in Superplasticizers in
Concrete, ACI SP-62, pp. 293–314 (Detroit, Michigan, 1979).

282
 5.42. A. MEYER, Experiences in the use of superplasticizers in Germany,
in Superplasticizers in Concrete, ACI SP-62, pp. 21–36 (Detroit, Michigan,
1979).
5.43. S. M. KHALIL and M. A. WARD, Influence of a lignin-based admixture on the
hydration of Portland cements, Cement and Concrete Research, 3, No. 6, pp.
677–88 (1973).
5.44. L. H. MCCURRICH, M. P. HARDMAN and S. A. LAMMIMAN, Chloride-free
accelerators, Concrete, 13, No. 3, pp. 29–32 (London, 1979).
5.45. P. SELIGMANN and N. R. GREENING, Studies of early hydration reactions of
portland cement by X-ray diffraction, Highway Research Record, No. 62, pp.
80–105 (Washington DC, 1964).
5.46. A. MEYER, Steigerung der Frühfestigkeit von Beton, 11 Cemento, 75, No. 3,
pp. 271–6 (1978).
5.47. V. M. MALHOTRA, Mechanical properties and durability of superplasticized
semi-lightweight concrete, CANMET Mineral Sciences Laboratory Report
MRP/MSL 79-131, 29 pp. (Canada Centre for Mineral and Energy
Technology, Ottawa, Sept. 1979).
5.48. CONCRETE SOCIETY, Admixtures for concrete, Technical Report TRCS 1, 12 pp.
(London, Dec. 1967).
5.49. F. P. GLASSER, Progress in the immobilization of radioactive wastes in
cement, Cement and Concrete Research, 22, Nos 2/3, pp. 201–16 (1992).
5.50. J. R. BIRCHALL and N. L. THOMAS, The mechanism of retardation of setting of
OPC by sugars, in The Chemistry and Chemically-Related Properties of
Cement, Ed. F. P. Glasser, British Ceramic Proceedings No. 35, pp. 305–315
(Stoke-on-Trent, 1984).
5.51. V. S. RAMACHANDRAN et al., The role of phosphonates in the hydration of
Portland cement, Materials and Structures, 26, No. 161, pp. 425–32 (1993).
5.52. ACI 212.4R-94, Guide for the use of high-range water-reducing admixtures
(superplasticizers) in concrete, in ACI Manual of Concrete Practice, Part 1:
Materials and General Properties of Concrete, 8 pp. (Detroit, Michigan,
1994).
5.53. B. MATHER, Chemical admixtures, in Concrete and Concrete-Making
Materials, Eds. P. Klieger and J. F. Lamond, ASTM Sp. Tech. Publ. No.
169C, pp. 491–9 (Detroit, Michigan, 1994).
5.54. D. L. KANTRO, Influence of water-reducing admixtures on properties of
cement paste – a miniature slump test, Research and Development Bulletin,
RD079.01T, Portland Cement Assn, 8 pp. (1981).
5.55. R. ASHWORTH, Some investigations into the use of sugar as an admixture to
concrete, Proc. Inst. Civ. Engrs, 31, pp. 129–45 (London, June 1965).
5.56. H. M. JONKERS and E. SCHLANGEN, Self-healing of cracked concrete: A
bacterial approach, Fracture Mechanics of Concrete and Concrete
Structures, 3, pp. 1821–1826 (2007).

283
Chapter 6. Strength of concrete

Strength of concrete is commonly considered its most valuable property, although, in

many practical cases, other characteristics, such as durability and permeability, may in
fact be more important. Nevertheless, strength usually gives an overall picture of the
quality of concrete because strength is directly related to the structure of the hydrated
cement paste. Moreover, the strength of concrete is almost invariably a vital element
of structural design and is specified for compliance purposes.
The mechanical strength of cement gel was discussed on p. 34; in this chapter
some empirical relations concerning the strength of concrete will be discussed.
Water/cement ratio
In engineering practice, the strength of concrete at a given age and cured in water at a
prescribed temperature is assumed to depend primarily on two factors only: the
water/cement ratio and the degree of compaction. The influence of air voids on
strength was discussed on p. 187, and at this stage we shall consider fully-compacted
concrete only: for mix proportioning purposes, this is taken to mean that the hardened
concrete contains about 1 per cent of air voids.
When concrete is fully compacted, its strength is taken to be inversely proportional
to the water/cement ratio. This relation was preceded by a so-called ‘law’, but really a
rule, established by Duff Abrams in 1919. He found strength to be equal to:

where w/c represents the water/cement ratio of the mix (originally taken by volume),
and K1 and K2 are empirical constants. The general form of the strength versus
water/cement ratio curve is shown in Fig. 6.1.

284
 Fig. 6.1. The relation between strength and water/cement ratio of concrete
Abrams’ rule, although established independently, is similar to a general rule
formulated by René Féret in 1896 in that they both relate strength of concrete to the
volumes of water and cement. Féret’s rule was in the form:

where fc is the strength of concrete, c, w and a are the absolute volumetric proportions
of cement, water, and air, respectively, and K is a constant.
It may be recalled that the water/cement ratio determines the porosity of the
hardened cement paste at any stage of hydration (see p. 30). Thus the water/cement
ratio and the degree of compaction both affect the volume of voids in concrete, and
this is why the volume of air in concrete is included in Féret’s expression.
The relation between strength and the volume of voids will be discussed more fully
in a later section. At this stage, we are concerned with the usual practical relation
between strength and the water/cement ratio. Figure 6.1 shows that the range of the
validity of the water/cement ratio rule is limited. At very low values of the
water/cement ratio, the curve ceases to be followed when full compaction is no longer
possible; the actual position of the point of departure depends on the means of
compaction available. It seems also that mixes with a very low water/cement ratio and
an extremely high cement content (probably above 530 kg/m3 (900 lb/yd3)) exhibit
retrogression of strength when large-size aggregate is used. Thus, at later ages, in this
type of mix, a lower water/cement ratio would not lead to a higher strength. This
behaviour may be due to stresses induced by shrinkage, whose restraint by aggregate

285
particles causes cracking of the cement paste or a loss of the cement–aggregate
bond.6.2
From time to time, the water/cement ratio rule has been criticized as not being
sufficiently fundamental. Nevertheless, in practice the water/cement ratio is the
largest single factor in the strength of fully compacted concrete. Perhaps the best
statement of the situation is that by Gilkey:6.74
“For a given cement and acceptable aggregates, the strength that may be developed
by a workable, properly placed mixture of cement, aggregate, and water (under the
same mixing, curing, and testing conditions) is influenced by the:
(a) ratio of cement to mixing water
(b) ratio of cement to aggregate
(c) grading, surface texture, shape, strength, and stiffness of aggregate particles
(d) maximum size of the aggregate.”
We can add that factors (b) to (d) are of lesser importance than factor (a) when
usual aggregates up to 40 mm ( in.) maximum size are employed. Those factors are,
nevertheless, present because, as pointed out by Walker and Bloem,6.74 “the strength of
concrete results from: (1) the strength of the mortar; (2) the bond between the mortar
and the coarse aggregate; and (3) the strength of the coarse aggregate particle, i.e. its
ability to resist the stresses applied to it”.
Figure 6.2 shows that the graph of strength versus water/cement ratio is
approximately in the shape of a hyperbola. This applies to concrete made with any
given type of aggregate and at any given age. It is a geometrical property of a
hyperbola y = k/x that y against 1/x plots as a straight line. Thus, the relation between
the strength and the cement/water ratio is approximately linear in the range of
cement/water ratios between about 1.2 and 2.5. This linear relationship, first
suggested in ref. 6.4, has been confirmed by Alexander and Ivanusec6.112 and by
Kakizaki et al.6.58 It is clearly more convenient to use than the water/cement ratio curve,
particularly when interpolation is desired. Figure 6.3 shows the data of Fig.
6.2 plotted with the cement/water ratio as abscissa. The values used apply to the given
cement only, and in any practical case the actual relation between strength and
cement/water ratio has to be determined.

286
 Fig. 6.2. Relation between 7-day strength and water/cement ratio for concrete
made with a rapid-hardening Portland cement

Fig. 6.3. A plot of strength against cement/water ratio for the data of Fig. 6.2

287
 The linearity of the relation between strength and cement/water ratio does not
extend beyond the cement/water ratio of 2.6, which corresponds to the water/cement
ratio of 0.38. In fact, for cement/water ratios larger than 2.6, there exists a different,
but still linear, relation with strength,6.59 as shown in Fig. 6.4. This figure represents
calculated values for cement pastes which have achieved maximum possible
hydration. For water/cement ratios smaller than 0.38, the maximum possible
hydration is less than 100 per cent (see p. 27); consequently, the slope of the curve is
different from that for higher values of the water/cement ratio. This observation is
worth remembering as nowadays mixes with water/cement ratios both somewhat
above and somewhat below 0.38 are often used.

Fig. 6.4. Relation between calculated strength of neat cement paste and
cement/water ratio. Maximum possible hydration is assumed to have taken place
(based on ref. 6.59)
The pattern of strength of high-alumina cement concrete is somewhat different
from that of concrete made with Portland cement, in that strength increases with the
cement/water ratio at a progressively decreasing rate.6.4
It must be admitted that the relations discussed here are not precise, and other
approximations can be made. For instance, it has been suggested that, as an
approximation, the relation between the logarithm of strength and the natural value of
the water/cement ratio can be assumed to be linear6.3 (cf. Abrams’ expression). As an
illustration, Fig. 6.5 gives the relative strength of mixes with different water/cement
ratios, taking the strength at the water/cement ratio of 0.4 as unity.

288
 Fig. 6.5. Relation between logarithm of strength and water/cement ratio6.3
Effective water in the mix
The practical relations discussed so far involve the quantity of water in the mix. This
needs a more careful definition. We consider as effective that water which occupies
space outside the aggregate particles when the gross volume of concrete becomes
stabilized, i.e. approximately at the time of setting. Hence the terms effective, free,
or net water/cement ratio.
Generally, water in concrete consists of that added to the mix and that held by the
aggregate at the time when it enters the mixers. A part of the latter water is absorbed
within the pore structure of the aggregate (see p. 129) while some exists as free water
on the surface of the aggregate and is therefore no different from the water added
direct into the mixer. Conversely, when the aggregate is not saturated and some of its
pores are therefore air-filled, a part of the water added to the mix will be absorbed by
the aggregate during the first half-hour or so after mixing. Under such circumstances
the demarcation between absorbed and free water is a little difficult.
On a site, the aggregate is as a rule wet, and the water in excess of that required for
the aggregate to be in a saturated and surface-dry condition is considered to be the
effective water of the mix. For this reason, the mix proportioning data are based
usually on the water in excess of that absorbed by the aggregate, that is the free water.
On the other hand, some laboratory tests refer to the total water added to a dry

289
aggregate. Care is, therefore, necessary in translating laboratory results into mix
proportions to be used on a site, and all reference to water/cement ratio must make it
clear if total rather than free water is considered.
Gel/space ratio
The influence of the water/cement ratio on strength does not truly constitute a law
because the water/cement ratio rule does not include many qualifications necessary
for its validity. In particular, strength at any water/cement ratio depends on: the
degree of hydration of cement and its chemical and physical properties; the
temperature at which hydration takes place; the air content of the concrete; and also
the change in the effective water/cement ratio and the formation of cracks due to
bleeding.6.5 The cement content of the mix and the properties of the aggregate–cement
paste interface are also relevant.
It is more correct, therefore, to relate strength to the concentration of the solid
products of hydration of cement in the space available for these products; in this
connection it may be relevant to refer again to Fig. 1.10. Powers6.6 has determined the
relation between the strength development and the gel/space ratio. This ratio is
defined as the ratio of the volume of the hydrated cement paste to the sum of the
volumes of the hydrated cement and of the capillary pores.
On p. 27, it was shown that cement hydrates to occupy more than twice its original
volume; in the following calculations the products of hydration of 1 ml of cement will
be assumed to occupy 2.06 ml; not all the hydrated material is gel, but as an
approximation we can consider it as such. Let
c = mass of cement
vc = specific volume of cement, that is volume of unit mass
wo = volume of mixing water, and
α = the fraction of cement that has hydrated.
Then, the volume of gel is 2.06cvcα, and the total space available to the gel
6.7

is cvcα + wo. Hence, the gel/space ratio is

Taking the specific volume of dry cement as 0.319 ml/g, the gel/space ratio
becomes:

The compressive strength of concrete tested by Powers6.7 was found to be

234r3 MPa (34 000r3 psi), and is independent of the age of the concrete or its mix
proportions. The actual relation between the compressive strength of mortar and the
gel/space ratio is shown in Fig. 6.6: it can be seen that strength is approximately
proportional to the cube of the gel/space ratio, and the figure 234 MPa (34 000 psi)
represents the intrinsic strength of gel for the type of cement and of specimen
used.6.8 Numerical values differ little for the usual range of Portland cements except
that a higher C3A content leads to a lower strength at a given gel/space ratio.6.5

290
 Fig. 6.6. Relation between the compressive strength of mortar and gel/space
ratio6.8
These calculations require a small modification to take account of the fact that the
specific gravity of the adsorbed water is 1.1 (see p. 37). Therefore, the actual volume
of voids is somewhat larger than assumed.
If the volume of air present in the cement paste is A, the ratio wo/c in the above
expression is replaced by (wo + A)/c (see Fig. 6.7). The resulting expression for
strength is similar to that of Féret but the ratio used here involves a
quantity proportional to the volume of hydrated cement instead of the total volume of
content, and is thus applicable at any age.

291
Fig. 6.7. Relation between the compressive strength of mortar and gel/space ratio,
modified to include entrapped air voids6.7
The expression relating strength to the gel/space ratio can be written in a number
of ways. It may be convenient to utilize the fact that the volume of non-evaporable
water, wn, is proportional to the volume of the gel; and also that the volume of mixing
water, wo, is related to the space available for the gel. The strength, fc, in pounds per
square inch, for fc greater than about 2000 psi, when the relation is approximately
linear, can then be written (using the original American units) in the form:6.6

Alternatively, the surface area of gel, Vm, can be used. Then (still in the American
units):

Figure 6.8 shows Powers’ actual data6.6 for cements with low C3A contents.

292
Fig. 6.8. Relation6.6 between the strength of cement paste and the ratio of surface
area of gel Vm to the volume of mixing water wo
The above expressions have been found to be valid for many cements but the
numerical coefficients may depend on the intrinsic strength of the gel produced by a
given cement. In other words, the strength of the cement paste depends primarily on
the physical structure of the gel but the effects of the chemical composition of cement
cannot be neglected; however, at later ages, these effects become minor only. Another
way of recognizing the properties of the gel is to say that strength depends primarily
on porosity but it is also affected by the ability of the material to resist crack
propagation, which is a function of bonding. Poor bond between two crystals can be
considered to be a crack.6.35
Porosity
The discussion in the preceding two sections showed that the strength of concrete is
fundamentally a function of the volume of voids in it. The relation between strength
and the total volume of voids is not a unique property of concrete but is found also in
other brittle materials in which water leaves behind pores: for instance, the strength of
plaster is also a direct function of its void content6.1 (see Fig. 6.9). Moreover, if the
strengths of different materials are expressed as a fraction of their respective strengths
at a zero porosity, a wide range of materials conform to the same relation between
relative strength and porosity, as shown in Fig. 6.10 for plaster, steel, iron,6.72 alumina
and zirconia.6.73 This general pattern is of interest in understanding the role of voids in
the strength of concrete. Moreover, the relation of Fig. 6.10 makes it clear why

293
cement compacts (see p. 281), which have a very low porosity, have a very high
strength.

Fig. 6.9. Strength of plaster as a function of its void content6.1

294
 Fig. 6.10. Influence of porosity on relative strength of various materials
Strictly speaking, strength of concrete is influenced by the volume of all voids in
concrete: entrapped air, capillary pores, gel pores, and entrained air, if present.6.10 An
example of the calculation of the total void content may be of interest and is given
below.
Let the given mix have proportions of cement, fine aggregate and coarse aggregate
of 1 : 3.4 : 4.2, with a water/cement ratio of 0.80. The entrapped air content has been
measured to be 2.3 per cent. Given that the specific gravity of the fine and coarse
aggregates is, respectively, 2.60 and 2.65, and on the assumption that the specific
gravity of cement is 3.15, the volumetric ratio of cement : fine aggregate : coarse
aggregate: water is

(1/3.15) : (3.4/2.60) : (4.2/2.65) : (0.80) = 0.318 : 1.31 : 1.58 : 0.80.

295
Because the air content is 2.3 per cent, the volume of the remaining materials must
add up to 97.7 per cent of the total volume of concrete. Thus, on a percentage basis,
the volumes are as follows:

We know that, in the given case, 0.7 of the cement has hydrated after 7 days of
curing in water (see, for instance, ref. 6.32). Therefore, continuing in percentage
volume units, we find the volume of the cement which has hydrated to be 5.5 and the
volume of unhydrated cement 2.3.
The volume of combined water is 0.23 of the mass of cement which has hydrated
(see p. 26), i.e. 0.23 × 5.5 × 3.15 = 4.0. On hydration, the volume of the solid products
of hydration becomes smaller than the sum of volumes of the constituent cement and
water by 0.254 of the volume of combined water (see p. 26). Hence, the volume of the
solid products of hydration is:

5.5 + (1 – 0.254) × 4.0 = 8.5.

Since the gel has a characteristic porosity of 28 per cent (see p. 26), the volume of
gel pores is wg such that wg/(8.5 + wg) = 0.28, whence the volume of gel pores is 3.3.
Thus, the volume of hydrated cement paste, inclusive of gel pores, is 8.5 + 3.3 = 11.8.
Now, the volume of dry cement which has hydrated and of mixing water is 5.5 +19.4
= 24.9. Hence, the volume of capillary pores is 24.9 – 11.8 = 13.1. Thus, the voids are:

The influence of the volume of pores on strength can be expressed by a power

function of the type:
fc = fc,0(1 – p)n
where p = porosity, that is, the volume of voids expressed as a fraction of the total
volume of concrete
fc = strength of concrete with porosity p
fc,0 = strength at zero porosity, and
n = a coefficient, which need not be constant.6.33
The exact form of the relation is, however, uncertain. Tests on pressed and heat-
treated cement compacts, as well as on ordinary cement paste, leave us still in doubt
as to whether the logarithm of porosity is linearly related to strength or to its
logarithm. Figures 6.11 and 6.12 illustrate the uncertainty. As far as the strength of
individual cement compounds is concerned, it has been found to be linearly related to
porosity (see Fig. 6.13).6.65

296
 Fig. 6.11. Relation between compressive strength and logarithm of porosity of
cement paste compacts for various treatments of pressure and high
temperature6.34

297
 Fig. 6.12. Relation between logarithm of compressive strength and logarithm of
porosity of cement paste compacts for various treatments of pressure and high
temperature (after ref. 6.34)

298
 Fig. 6.13. Relation between compressive strength and porosity of pure
compounds6.65
In addition to their volume, the shape and size of pores are also factors. The shape
of the solid particles and their modulus of elasticity also influence the stress
distribution and, therefore, stress concentration, within concrete. An example of pore
distribution in concrete is shown in Fig. 6.14.6.68 Similar results were obtained by
Hearn and Hooton.6.113

Fig. 6.14. Cumulative volume of pores larger than indicated pore diameter in
concrete with a water/cement ratio of 0.45 at 20 °C (based on ref. 6.68)
The effect of porosity on the strength of hydrated cement paste has been studied
widely. Care is required in translating observations on laboratory-made specimens of
neat cement paste into usable information about concrete, but an understanding of the
effect of porosity on strength of hydrated cement paste is valuable.
There is no doubt that porosity defined as the total volume of the overall volume of
pores larger than gel pores, expressed as a percentage of the overall volume of the
hydrated cement paste, is a primary factor influencing the strength of the cement paste.
A linear relation between strength and porosity, within the range of the latter between
5 and 28 per cent, was established by Rössler and Odler.6.63 The effect of pores smaller
than 20 nm in diameter was found to be negligible.6.64 The relation between the
strength of mortar and porosity based on volume of pores larger than 20 nm in
diameter is shown in Fig. 6.15.6.66 Consequently, in addition to total porosity, the effect
of pore size distribution on strength must be considered. Generally, at a given porosity,
smaller pores lead to a higher strength of the cement paste.

299
 Fig. 6.15. Relation between compressive strength of mortar and porosity
calculated from the volume of pores larger than 20 nm in diameter (based on
ref. 6.66)
Although the pore size is, for convenience, expressed as a diameter, all the pores
are by no means cylindrical or spherical in shape: the ‘diameter’ represents a sphere
with the same ratio of volume to surface area as the totality of pores. It is only
macropores, that is those whose diameter is larger than about 100 nm, that are
approximately spherical. Figure 6.16 shows a diagrammatic representation of the
various pores. This figure is an extension and modification of Fig. 1.13. The spherical
pores originate from residual air bubbles or from imperfect cement particle packing
but are not readily detected in porosimetry measurements because they are accessible
only through connecting pores which have a narrow entrance (see Fig. 6.16).6.70

300
 Fig. 6.16. Diagrammatic representation of the pore system in hydrated cement
paste (based on Rahman’s model in ref. 6.70)
The dependence of the strength of hydrated cement paste on its porosity and on the
pore size distribution is fundamental. Research papers occasionally consider a relation
between strength and the gypsum content of the cement, but this is the outcome of the
fact that the gypsum content affects the progress of hydration of cement and thus the
pore distribution within the hydrated cement paste. The problem is, however,
complicated by the fact that different methods of determination of porosity do not
always yield the same values.6.69 The main reason for this is that the process of
porosimetry measurement, especially if it involves removal or addition of water,
affects the structure of the hydrated cement paste.6.67 The use of mercury intrusion in
studies of the pore system in the cement paste is discussed by Cook and
Hover.6.114 This method assumes that pores become narrower with depth while, in fact,
some pores have a constricted entrance; this distorts the value of porosity measured
by mercury intrusion porosimetry.6.115
As pointed out earlier, most of the experimental work on porosity of hydrated
cement paste has been performed on specimens of neat cement paste or of mortar. In
concrete, the pore characteristics of the hydrated cement are somewhat different
because of the influence of coarse aggregate particles on the cement paste in their
neighbourhood. Winslow and Liu6.68 found that, with the same paste composition and
at the same degree of hydration, the presence of coarse aggregate results in an
increased porosity; even the presence of fine aggregate has a similar, but smaller,
effect. The difference in porosity between concrete and neat cement paste, at the same
water/cement ratio, increases with the progress of hydration and arises from the
presence in concrete of some pores larger than those which can exist in neat cement
paste.
Cement compacts
Cement compacts are manufactured by application of very high pressure with
simultaneous high temperature. They do not, therefore, come under the heading of
concrete, but are of interest in elucidating the role of porosity in strength because a
porosity as low as 1 per cent can be achieved.6.34
One of the strongest cement-based materials reported6.62 to have been produced had
a water/cement ratio of 0.08: when compacted, it had a strength of 345 MPa (50 000
psi). Application of pressure of 340 MPa (49 500 psi) and temperature of 250 °C
(480 °F) has resulted in compacts with a compressive strength of about 660 MPa (95
000 psi) and a tensile splitting strength of 64 MPa (9300 psi).6.34
Extrapolation of an experimental relation between porosity and compressive
strength of specimens of the individual compounds in Portland cement, at a
water/solid ratio of 0.45, suggests that, at zero porosity, the strength is about 500
MPa.6.65 This can be compared with the value calculated by Nielsen6.59 who estimates
the strength of hydrated cement paste at zero porosity to be 450 MPa.
These values, although not unique, represent the intrinsic strength of hardened
Portland cement paste.
Influence of properties of coarse aggregate on strength
Although the relation between strength and the water/cement ratio is generally valid,
it is not independent of other factors. One of these is discussed in this section.

301
 Vertical cracking in a specimen subjected to uniaxial compression starts under a
load equal to 50 to 75 per cent of the ultimate load. This has been determined from
measurements of the velocity of sound transmitted through the concrete,6.22 and also
using ultrasonic pulse velocity techniques.6.23 The stress at which the cracks develop
depends largely on the properties of the coarse aggregate: smooth gravel leads to
cracking at lower stresses than rough and angular crushed rock, probably because
mechanical bond is influenced by the surface properties and, to a certain degree, by
the shape of the coarse aggregate.6.19
The properties of aggregate affect thus the cracking load, as distinct from the
ultimate load, in compression and the flexural strength in the same manner, so that the
relation between the two quantities is independent of the type of aggregate
used. Figure 6.17 shows Jones and Kaplan’s6.19 results, each symbol representing a
different type of coarse aggregate. On the other hand, the relation between the flexural
and compressive strengths depends on the type of coarse aggregate used (see Fig.
6.18) because (except in high strength concrete) the properties of aggregate,
especially its shape and surface texture, affect the ultimate strength in compression
very much less than the strength in tension or the cracking load in compression. This
behaviour was confirmed by Knab.6.71 In experimental concrete, entirely smooth coarse
aggregate led to a lower compressive strength, typically by 10 per cent, than when
roughened.6.38

302
Fig. 6.17. Relation between flexural strength and compressive stress at cracking
for concretes made with different coarse aggregates6.19 (Crown copyright)

303
Fig. 6.18. Relation between compressive strength and indirect tensile strength for
concretes of constant workability made with various aggregates (water/cement
ratio between 0.33 and 0.68, aggregate/cement ratio between 2.8 and
10.1)6.39 (Crown copyright)
The influence of the type of coarse aggregate on the strength of concrete varies in
magnitude and depends on the water/cement ratio of the mix. For water/cement ratios
below 0.4, the use of crushed aggregate has resulted in strengths up to 38 per cent
higher than when gravel is used. The behaviour at a water/cement ratio of 0.5 is
shown in Fig. 6.19.6.39 With an increase in the water/cement ratio, the influence of
aggregate falls off, presumably because the strength of the hydrated cement paste
itself becomes paramount and, at a water/cement ratio of 0.65, no difference in the
strengths of concretes made with crushed rock and gravel has been observed.6.24

304
 Fig. 6.19. Relation between compressive strength and age for concretes made
with various aggregates (water/cement ratio = 0.5)6.39 (Crown copyright)
The influence of aggregate on flexural strength seems to depend also on the
moisture condition of the concrete at the time of test.6.60
The shape and surface texture of coarse aggregate affect also the impact strength of
concrete, the influence being qualitatively the same as on the flexural strength6.61 (see
p. 117).
Kaplan6.25 observed that the flexural strength of concrete is generally lower than the
flexural strength of corresponding mortar. Mortar would thus seem to set the upper
limit to the flexural strength of concrete and the presence of the coarse aggregate
generally reduces this strength. On the other hand, the compressive strength of
concrete is higher than that of mortar, which, according to Kaplan, indicates that the
mechanical interlocking of the coarse aggregate contributes to the strength of concrete
in compression. This behaviour has not, however, been confirmed to apply generally,
and the question of the influence of aggregate on strength is considered further in the
next section. At this stage, it is useful to note that coarse aggregate particles act as
crack arresters so that, under an increasing load, another crack is likely to open.
Failure is, therefore, gradual and, even in tension, there exists a descending part of the
stress–strain curve.
Influence of aggregate/cement ratio on strength
The anomalous behaviour of extremely rich mixes with respect to strength was
mentioned on p. 270, but the richness of the mix affects the strength of all medium-

305
and high-strength concretes, i.e. those with a strength of about 35 MPa (5000 psi) or
more. There is no doubt that the aggregate/cement ratio, is only a secondary factor in
the strength of concrete but it has been found that, for a constant water/cement ratio, a
leaner mix leads to a higher strength6.12 (see Fig. 6.20).

Fig. 6.20. Influence of the aggregate/cement ratio on strength of concrete6.13

The reasons for this behaviour are not clear. In certain cases, some water may be
absorbed by the aggregate: a larger amount of aggregate absorbs a greater quantity of
water, the effective water/cement ratio being thus reduced. In other cases, a higher
aggregate content would lead to lower shrinkage and lower bleeding, and therefore to
less damage to the bond between the aggregate and the cement paste; likewise, the
thermal changes caused by the heat of hydration of cement would be smaller.6.80 The
most likely explanation, however, lies in the fact that the total water content per cubic
metre of concrete is lower in a leaner mix than in a rich one. As a result, in a leaner
mix, the voids form a smaller fraction of the total volume of concrete, and it is these
voids that have an adverse effect on strength.
Studies on the influence of aggregate content on the strength of concrete with a
given quality of cement paste indicate that, when the volume of aggregate (as a
percentage of the total volume) is increased from zero to 20, there is a gradual
decrease in compressive strength, but between 40 and 80 per cent there is an
increase.6.40 The pattern of behaviour is shown in Fig. 6.21. The reasons for this effect
are not clear, but it is the same at various water/cement ratios.6.41 The influence of the
volume of aggregate on tensile strength is broadly similar6.40 (Fig. 6.22).

306
 Fig. 6.21. Relation6.40 between the compressive strength of cylinders (100 mm
diameter, 300 mm in length) and volume of aggregate at a constant water/cement
ratio of 0.50

307
Fig. 6.22. Relation6.40 between direct tensile strength and volume of aggregate at a
constant water/cement ratio of 0.50
These effects are smaller in cubes than in cylinders or prisms. In consequence, the
ratio of cylinder strength to cube strength (cf. p. 596) decreases as the volume of
aggregate increases from zero to 40 per cent.6.45 The explanation lies probably in the
greater influence of the aggregate on the crack pattern when the end effect of platens
is absent (see p. 587).
Nature of strength of concrete
The paramount influence of voids in concrete on its strength has been repeatedly
mentioned, and it should be possible to relate this factor to the actual mechanism of
failure. For this purpose, concrete is considered to be a brittle material, even though it
exhibits a small amount of plastic action, as fracture under static loading takes place
at a moderately low total strain; a strain of 0.001 to 0.005 at failure has been
suggested as the limit of brittle behaviour. High strength concrete is more brittle than
normal strength concrete but there is no quantitative method of expressing the
brittleness of concrete whose behaviour in practice falls between the brittle and the
ductile types.
Strength in tension
The actual (technical) strength of hydrated cement paste or of similar brittle materials
such as stone is very much lower than the theoretical strength estimated on the basis
of molecular cohesion, and calculated from the surface energy of a solid assumed to
be perfectly homogeneous and flawless. The theoretical strength has been estimated to
be as high as 10.5 GPa (1.5 × 106 psi).
This discrepancy can be explained by the presence of flaws postulated by
Griffith.6.17 These flaws lead to high stress concentrations in the material under load so

308
that a very high stress is reached in very small volumes of the specimen with a
consequent microscopic fracture, while the average (nominal) stress in the whole
specimen is comparatively low. The flaws vary in size, and it is only the few largest
ones that cause failure: the strength of a specimen is thus a problem of statistical
probability, and the size of the specimen affects the probable nominal stress at which
failure is observed.
Hydrated cement paste is known to contain numerous discontinuities – pores,
microcracks and voids – but the exact mechanism through which they affect
the strength is not known. The voids themselves need not act as flaws, but the flaws
may be cracks in individual crystals associated with the voids6.14 or caused by
shrinkage or poor bond. This situation is not surprising in view of the heterogeneous
nature of concrete and of the method of combining the various phases of this
composite material into a single whole. Alford et al.6.81 confirmed that the pores in the
cement paste are not the only possible critical flaws. In unsegregated concrete, the
voids are distributed in a random manner,6.15 which is a condition necessary for the
application of Griffith’s hypothesis. While we do not know the exact mechanism of
rupture of concrete, this is probably related to the bond within the hydrated cement
paste and between the paste and the aggregate.
Griffith’s hypothesis postulates microscopic failure at the location of a flaw, and it
is usually assumed that the ‘volume unit’ containing the weakest flaw determines the
strength of the entire specimen. This statement implies that any crack will spread
throughout the section of the specimen subjected to the given stress or, in other words,
an event taking place in an element is identified with the same event taking place in
the body as a whole.
This behaviour can be met with only under a uniform stress distribution, with the
additional proviso that the ‘second weakest’ flaw is not strong enough to resist a stress
of n/(n – 1) times the stress at which the weakest flaw failed, where n is the number of
elements in the section under load, each element containing one flaw.
Whereas local fracture starts at a point and is governed by the conditions at that
point, the knowledge of stress at the most highly stressed point in the body is not
sufficient to predict failure. It is necessary to know also the stress distribution in a
volume sufficiently extended round this point because the deformational response
within the material, particularly near failure, depends on the behaviour and state of the
material surrounding the critical point, and the possibility of spreading of failure is
strongly affected by this state. This would explain, for instance, why the maximum
fibre stresses in flexure specimens at the instant of incipient failure are higher than the
strength determined in uniform direct tension: in the latter case, the propagation of
fracture is not blocked by the surrounding material. Some actual data on the relation
between the strength in flexure and in splitting tension are given in Fig. 12.8.
We can see then that, in a given specimen, different stresses will produce fracture
at different points, but it is not possible physically to test the strength of an individual
element without altering its condition in relation to the rest of the body. If the strength
of a specimen is governed by the weakest element in it, the problem becomes that of
the proverbial weakest link in a chain. In statistical terms, we have to determine the
least value (i.e. the strength of the most effective flaw) in a sample of size n,
where n is the number of flaws in the specimen. The chain analogy may not be quite
correct because, in concrete, the links may be arranged in parallel as well as in series,
but computations on the basis of the weakest link assumption yield results of the
correct order. It follows that the strength of a brittle material such as concrete cannot

309
be described by an average value only: an indication of the variability of strength
must be given, as well as information about the size and shape of the specimens.
These factors are discussed in Chapter 12.
Cracking and failure in compression
Griffith’s hypothesis applies to failure under the action of a tensile force but it can be
extended to fracture under bi- and triaxial stress and also under uniaxial compression.
Even when two principal stresses are compressive, the stress along the edge of the
flaw is tensile at some points, so that fracture can take place. Orowan6.16 calculated the
maximum tensile stress at the tip of the flaw of the most dangerous orientation
relative to the principal stress axes as a function of the two principal stresses P and Q.
The fracture criteria are represented graphically in Fig. 6.23, where K is the tensile
strength in direct tension. Fracture occurs under a combination of P and Q such that
the point representing the state of stress crosses the curve outwards onto the shaded
side.

Fig. 6.23. Orowan’s criteria of fracture under biaxial stress6.16

From Fig. 6.23, it can be seen that fracture can occur when uniaxial compression is
applied; this has in fact been observed in tests on concrete compression test
specimens.6.18 The nominal compressive strength in this case is 8K, i.e. 8 times the
tensile strength determined in a direct tension test. This figure is in good agreement
with the observed values of the ratio of the compressive to tensile strengths of
concrete. There are, however, difficulties in reconciling certain aspects of Griffith’s
hypothesis with the observed direction of cracks in compression specimens. It is
possible, though, that failure in such a specimen is governed by the lateral strain
induced by Poisson’s ratio. The value of Poisson’s ratio for concrete is such that, for

310
elements sufficiently removed from the platens of the testing machine, the resulting
lateral strain can exceed the ultimate tensile strain of concrete. Failure occurs then by
splitting at right angles to the direction of the load, as in the splitting test (see p. 600),
and this has been frequently observed, especially in specimens whose height is greater
than their breadth.6.18 The view that concrete fails by tensile splitting when subjected to
uniaxial or biaxial compression was confirmed by Yin et al.6.86
There are strong indications, first developed in ref. 6.14, that it is not a limiting
strain but a limiting tensile strain that determines the strength of concrete under static
loading: this is usually assumed to be between 100 × 10–6 and 200 × 10–6. The failure
criterion of limiting tensile strain is supported by an analysis advanced by Lowe.6.36 It
has been found that, at the point of initial cracking, the strain on the tension face of a
beam in flexure and the lateral tensile strain in a cylinder in uniaxial compression are
of similar magnitude.6.21 The tensile strain in a beam at cracking is:

where E is the modulus of elasticity of concrete over the linear range of deformation.
Now, the lateral strain in a compression specimen when cracking is first observed is:

where μ is the static Poisson’s ratio, and E is the same as above. From the observed
equality of the two strains it would appear that:

Poisson’s ratio varies generally between about 0.15 for high strength concrete and
0.22 for low strength concrete (see p. 421), and it is significant that the ratio of the
nominal tensile and compressive strengths for different concretes varies in a similar
manner and between approximately the same limits. There is thus a possible
connection between the ratio of nominal strengths and Poisson’s ratio, and there are
good grounds for suggesting that the mechanism producing the initial cracks in
uniaxial compression and in flexure tension is the same.6.19 The nature of this
mechanism has not been established. It is probable that cracking is due to local
breakdowns in bond between the cement and the aggregate.6.20 However, the basic
mechanism of compressive failure of concrete has not been reliably established and
even the definition of failure of concrete is not obvious. One view is to associate
failure with the so-called discontinuity point, defined as the point at which the
volumetric strain stops decreasing and Poisson’s ratio starts to increase
sharply.6.52,6.53 At this stage, extensive mortar cracking starts to develop (see p. 300).
This is the beginning of instability, and sustained loading above this point will lead to
failure. The lateral tensile strain at discontinuity depends on the level of axial
compression and is greater for stronger concrete; Carino and Slate6.53 observed an
average value of about 300 × 10–6 at a stress of 7.5 MPa (1100 psi). It should be noted,
however, that other workers6.119 reported that hydrated cement paste is damaged
progressively and without the discontinuity point being a significant feature.
The ultimate failure under the action of a uniaxial compression is either a tensile
failure of cement crystals or of bond in a direction perpendicular to the applied load,
or is a collapse caused by the development of inclined shear planes.6.20 It is probable

311
that ultimate strain is the criterion of failure, but the level of strain varies with the
strength of concrete: the higher the strength the lower the ultimate strain. While actual
values depend on the method of test, typical values are as given in Table 6.1.

Table 6.1. Typical Values of Compressive Strain at Failure

Failure under multiaxial stress

Under triaxial compression, when the lateral stresses are high, failure must take place
by crushing: the mechanism is, therefore, different from that described above, the
behaviour of concrete changing from brittle to ductile. An increase in lateral
compression increases the axial load that can be sustained, as shown, for instance,
in Fig. 6.24.6.26 With very high lateral stresses, extremely high strengths have been
recorded6.11 (Fig. 6.25). It should be noted that, if the development of pore water
pressure in concrete is limited by allowing the displaced pore water to escape through
the loading platens, then the apparent strength is higher.6.75 Thus, in practice, a possible
development of pore pressure is of importance.6.84

312
 Fig. 6.24. Influence of lateral stress on the axial stress at failure of neat cement
paste and of mortar6.26

313
 Fig. 6.25. Influence of high lateral stress on the axial stress at failure of
concrete6.11
A confining lateral stress of 520 MPa has been reported to lead to an axial stress of
1200 MPa.6.82 If the lateral compressive stress increased progressively with the
increase in the axial stress, even higher values of the axial stress can be reached: 2080
MPa has been reached, coupled with a large decrease in porosity.6.82
A lateral tensile stress has a similar influence but, of course, in the opposite
direction.6.11 This behaviour agrees well with the theoretical considerations on the
previous page.
In practice, failure of concrete takes place over a range of stresses rather than as an
instantaneous phenomenon, so that ultimate failure is a function of the type of
loading.6.19 This is of especial interest when repeated loading is applied – a condition

314
frequently met with in practice. Fatigue strength of concrete is considered in Chapter
7.
A general biaxial stress interaction curve is shown in Fig. 6.26.6.78 A large
interaction is observed when there is a considerable frictional restraint at the platens
but, when the end restraint of the specimen is effectively eliminated (e.g. by the use of
steel brush platens, see p. 589), the effect is much smaller. It can be seen from Fig.
6.26, that, under a biaxial stress σ1 = σ3, the strength is only 16 per cent higher than in
uniaxial compression; biaxial tensile strength is no different from uniaxial tensile
strength.6.78 These findings were confirmed by other workers.6.9,6.54,6.86 However, some
differences were observed due to the variation in the rate of loading and the type of
coarse aggregate in the concrete.6.86 Experimental data on interaction are plotted in Fig.
6.27; these were obtained with steel brush platen loading and by the use of fluid
membranes and solid platens.6.46 Some contradictory data of other investigators can be
explained by the use of uncertain end restraints.

Fig. 6.26. Interaction curve for biaxial stress when the end restraint is effectively
eliminated6.78 (σ1 and σ3 are the biaxial stresses applied)

315
 Fig. 6.27. Strength of concrete under multiaxial stress as measured by various
investigators. Wet or air-dried concrete6.46 (fc = compressive strength)
The level of uniaxial compressive strength virtually does not affect the shape of the
curve or the magnitude of the values given by it;6.78 the prism strength range tested was
19 to 58 MPa (2700 to 8350 psi) and both the water/cement ratio and cement content
varied widely. However, in compression–tension and in biaxial tension, the relative
strength at any particular biaxial stress combination decreases as the level of uniaxial
compressive strength increases.6.78 This accords with the general observation that the
ratio of uniaxial tensile strength to uniaxial compressive strength decreases as the
compressive strength level rises (see p. 311); in these tests, the ratio was 0.11, 0.09
and 0.08 at a uniaxial compressive strength level of 19, 31 and 58 MPa (2700, 4450
and 8350 psi) respectively.6.78
Generally, triaxial compression increases the strength of weaker or leaner concrete
relatively more than that of stronger or richer concrete.6.47 For the range of
conventional concretes, Hobbs6.47 found that, under triaxial compression, the major
principal stress at failure, σ1, can be expressed, on average, as:

where σ3 = minor principal stress, and

fcyl = cylinder strength.

316
 The limited information on lightweight aggregate concrete suggests that the
influence of σ3 is not as large as with normal aggregates;6.46 therefore, the coefficient
4.8 in the above equation can be reduced to about 3.2.
The combined strength results for concretes in triaxial compression and in biaxial
compression plus tension, may be represented6.47 by the equation:

all values being averages in MPa, and compression being taken as positive.
The values given in Eqs (2) and (3) apply to conventional concretes only, and not
to neat cement pastes or mortars.
Substituting equations (2) and (3) into equation (1), but using the lower bound, and
not average, values, yields the failure criterion for conventional concretes:

This equation is plotted in Fig. 6.28 for various values of the cylinder strength, fcyl.
The generality of this equation should not be overestimated because, as
Hobbs6.47 points out, the tensile strength and compressive strength of concrete are not
equally affected by the aggregate type and grading and by the direction of the applied
stress relative to the direction of casting. In each case, the tensile strength is more
sensitive. It should also be noted that the intermediate principal stress, σ2, affects the
value of σ1.6.85

317
 Fig. 6.28. Failure stresses in concrete under biaxial stress6.47
The preceding discussion has shown that, while strength of concrete is an inherent
property of the material, as measured in practice it is also a function of the stress
system which is acting. Mather6.77 pointed out that, ideally, it should be possible to
express the failure criteria under all possible stress combinations by a single stress
parameter, such as strength in uniaxial tension. However, such a solution has not yet
been found.
Berg6.56 developed an equation of strength for concrete whose parameters are: the
stress at the initiation of crack propagation, the splitting (tensile) strength, and the
uniaxial compressive strength. This equation can be used for an analytical evaluation
of the failure of concrete under combined states of stress but it ceases to apply when
the tensile strength cannot be reached. Other approaches6.79 also have somewhat
limited validity.
Full understanding of the failure behaviour of concrete requires consideration of
fracture energy, that is the energy absorbed in a unit area of crack surface. This is a
subject of study of fracture mechanics, dealt with in specialist publications, e.g.
refs 6.87 and 6.88. However, fracture mechanics has so far not succeeded in
developing material parameters which can adequately quantify the resistance of
concrete to cracking.
Microcracking

318
Because failure of concrete is the consequence of cracking, it is useful to consider this
topic in some detail. In this section, only microcracking is considered. More general
aspects of cracking will be discussed in Chapter 10 as this requires a prior
consideration of the stress–strain relation of concrete.
Investigations have shown that very fine cracks at the interface between coarse
aggregate and cement paste exist, in fact, even prior to application of the load on
concrete.6.76 They are probably due to the inevitable differences in mechanical
properties between the coarse aggregate and the hydrated cement paste, coupled with
shrinkage or thermal movement. Microcracking has been observed not only in normal
strength concrete but also in wet-cured concrete with a water/cement ratio as low as
0.25, which had never been subjected to loading.6.92 According to Slate and
Hover,6.91 pre-loading microcracks are largely responsible for the low tensile strength
of concrete.
Microcracks have not been universally defined in terms of size, but an upper limit
of 0.1 mm has been suggested;6.91 this is the smallest size which can typically be
detected by the naked eye. For engineering purposes, a lower limit can be taken as the
smallest crack which can be observed using an optical microscope. As an increasing
load is being applied, these microcracks remain stable up to about 30 per cent, or
more, of the ultimate load and then begin to increase in length, width, and number.
The overall stress under which they develop is sensitive to the water/cement ratio of
the paste. This is the stage of slow crack propagation.
Upon further increase in load, up to between 70 and 90 per cent of the ultimate
strength, cracks open through the mortar (cement paste and fine aggregate); they
bridge the bond cracks so that a continuous crack pattern is formed.6.76 This is the fast
crack propagation stage. The stress level at the onset of this stage is higher in high
strength concrete than in normal concrete.6.90 The increase in the cumulative length of
the microcracks is large; this was measured using neutron radiography.6.116 However,
high strength concrete exhibits a lower cumulative length of microcracks than normal
strength concrete.6.90
The onset of the fast crack propagation stage corresponds to the discontinuity point
in the volumetric strain (referred to on p. 422). If the load is sustained, failure may
take place with time. This occurs both in normal strength and in high strength
concrete.6.90
Interesting results of measurement of crack length are shown in Fig. 6.29.6.37 It can
be seen that there was very little increase in the total length between the beginning of
loading and a stress equal to about 0.85 of the prism strength.6.37 A further increase in
stress resulted in a large increase in the total length of cracks. At a stress/strength ratio
of about 0.95, not only interface (bond) cracks but also mortar cracks were present,
and many cracks tended to become oriented roughly parallel to the direction of the
applied load. Once the specimen reached the descending part of the stress/strain curve
the rate of increase in the crack length and width became large.

319
 Fig. 6.29. Relation between the observed length of cracks in an area of 100
mm2 and the stress/strength ratio in compression (based on prisms)6.37
Figure 6.29 also shows the crack development under a cyclic stress alternating
between zero and 0.85 of the prism strength. Immediately prior to failure, the cracks
became longer and wider. Likewise, sustained loading at a stress/strength ratio of 0.85
led to an increase in cracking prior to failure.6.37
The preceding discussion has shown that microcracking is a general feature of
concrete. As long as the cracks are stable, their presence is not harmful. Paradoxically,
while the interface between coarse aggregate and the hydrated cement paste is the
locus of early microcracks, it is the presence of coarse aggregate particles that
prevents the opening of a single wide crack: these particles act as microcrack arrestors.
The heterogeneity of concrete is thus beneficial. The aggregate–paste bond surfaces
form all the possible angles with the direction of the external force. As a result, the
local stress varies substantially above and below the nominal applied stress. The
aggregate–paste interface will be discussed in the next section.
The existence of submicrocracks, defined as cracks which can be detected using a
scanning electron microscope at a magnification of at least 1250, has been
reported.6.111 This is not surprising because, in concrete, there are discontinuities at any
level, however small. There is no evidence, however, that submicrocracks are a factor
in the strength of concrete.

320
Aggregate–cement paste interface
The observation that microcracking is initiated at the interface between coarse
aggregate and the surrounding mortar and that, at failure, the crack pattern includes
the interface, points to the importance of this part of the concrete. It is therefore
necessary to understand the properties and behaviour of the interface zone, sometimes
called the transition zone.
The first fact to note is that the microstructure of the hydrated cement paste in the
immediate vicinity of coarse aggregate particles differs from that of the bulk of the
cement paste. The main reason for this is that, during mixing, dry cement particles are
unable to become closely packed against the relatively large particles of aggregate.
This situation is similar to the ‘wall-effect’ at the surface of cast concrete surfaces
(see p. 611) although on a much smaller scale. There is thus less cement present to
hydrate and fill the original voids. In consequence, the interface zone has a much
higher porosity than the hydrated cement paste further away from the coarse
aggregate6.94 (see Fig. 6.30). The influence of porosity on strength, discussed earlier in
this chapter, explains the weakness of the interface zone.

Fig. 6.30. Variation in porosity of hydrated cement paste with distance from the
surface of an aggregate particle (based on ref. 6.94)

321
 The microstructure of the interface zone is as follows. The surface of the aggregate
is covered with a layer of oriented crystalline Ca(OH)2, about 0.5 μm thick, behind
which there is a layer of C-S-H of about the same thickness. This is referred to as a
duplex film. Moving further away from the aggregate, there is the main interface zone,
some 50 μm thick, containing products of hydration of cement with larger crystals of
Ca(OH)2 but without any unhydrated cement.6.57
The significance of the above distribution is twofold. First, the complete hydration
of cement indicates that the water/cement ratio at the interface is higher than
elsewhere. Second, the presence of large crystals of Ca(OH)2 indicates that the
porosity at the interface is higher than elsewhere; this confirms the ‘wall effect’
referred to earlier.
The strength of the interface zone can increase with time in consequence of a
secondary reaction between the Ca(OH)2 present there and pozzolana. Silica fume,
which is very much finer than cement particles, is particularly effective. This topic is
discussed in Chapter 13.
Although the interface zone of primary interest is that at the surface of coarse
aggregate particles, such a zone is also formed around the fine aggregate
particles.6.93 Here, the thickness of the interface zone is smaller but the surface
effects originating from the fine particles interfere with those of the coarse aggregate
and thus affect the overall extent of the interface zone.6.93
The mineralogical characteristics of the fine aggregate affect the microstructure of
the transition zone: in the case of limestone, there is chemical reaction between the
limestone and the cement paste and, consequently, a dense interface zone is formed.6.95
As far as lightweight aggregate is concerned, if it has a dense outer layer, then the
situation at the interface is the same as with normal weight aggregate.6.89 However,
lightweight aggregate with a more porous outer layer, which encourages the migration
of mobile ions towards it,6.96 leads to the formation of a more dense interface zone and
also to improved mechanical interlocking of the aggregate particles and the hydrated
cement paste.6.89
The study of the interface zone in actual concrete is difficult. Consequently,
experiments on the interface between a single rock particle and cement paste have
been resorted to. However, the results of such tests may be misleading as they do not
include the effects of the interference of other coarse aggregate particles6.94 or even of
fine aggregate. Moreover, the laboratory-made artefact of a single particle covered by
cement paste has not undergone the process of mixing, in which shearing action
influences the microstructure of the cement paste at the time of setting. Furthermore,
in actual concrete, bleeding may result in water-filled voids on the underside of coarse
aggregate particles, and it is at this type of interface that massive crystals of
Ca(OH)2 have been observed. More generally, the interface between the cement paste
and the coarse aggregate is a zone of stress concentrations arising from the difference
in the modulus of elasticity and the Poisson’s ratio of the two materials.
Effect of age on strength of concrete
The relation between the water/cement ratio and the strength of concrete applies to
one type of cement and one age only, and also assumes wet-curing conditions. On the
other hand, the strength versus gel/space ratio relationship has a more general
application because the amount of gel present in the cement paste at any time is itself
a function of age and type of cement. The latter relation thus allows for the fact that

322
different cements require a different length of time to produce the same quantity of
gel.
The rate of gain of strength of different cements was discussed in Chapter 2,
and Figs 2.1 and 2.2 show typical strength–time curves. The influence of the curing
conditions on the development of strength is considered in Chapter 7, but here we are
concerned with the practical problem of strength of concrete at different ages.
In concrete practice, the strength of concrete is traditionally characterized by the
28-day value, and some other properties of concrete are often referred to the 28-day
strength. There is no scientific significance in the choice of the age of 28 days; it is
simply that early cements gained strength slowly and it was necessary to base the
strength description on concrete in which a significant hydration of cement had
already taken place. The specific choice of a multiple of weeks was, in all likelihood,
made so that testing, like placing, would fall on a working day. In modern Portland
cements, the rate of hydration is much greater than in the past, both because they have
a much higher fineness and because they have a higher C3S content. This is, however,
not necessarily the case with all blended cements.
It is arguable that a shorter period than 28 days could be used for the
characterization of strength, but the age of 28 days seems to have acquired an
immutable position. Thus, compliance with the specification is almost invariably laid
down in terms of the 28-day strength. If, for some reason, the 28-day strength is to be
estimated from the strength determined at an earlier age, say 7 days, then the relation
between the 28-day and the 7-day strengths has to be established experimentally for
the given mix. For this reason, the various expressions for the ratio of the two
strengths are no longer thought to be reliable, and they will not be discussed. The
consequences of the change in the strength-development characteristics which took
place in the 1970s will be discussed on p. 336.
Not only the properties of cement but the water/cement ratio also affect the rate of
gain of strength of concrete. Mixes with a low water/cement ratio gain strength,
expressed as a percentage of long-term strength, more rapidly than mixes with higher
water/cement ratios6.83 (Fig. 6.31). This is because in the former case the cement grains
are closer to one another and a continuous system of gel is established more rapidly. It
should be noted that in a hot climate the early strength gain is high and the ratio of the
28-day to 7-day strengths tends to be lower than in cooler weather. This is also the
case with some lightweight aggregate concretes.

323
 Fig. 6.31. Relative gain of strength with time in concretes with different
water/cement ratios, made with ordinary Portland cement6.83
Knowledge of the strength–time relation is of importance when a structure is to be
put into use, that is, subjected to full loading, at a later age: in such a case, the gain in
the strength after the age of 28 days can be taken into account in design. In some other
situations, for example in precast or in prestressed concrete, or when early removal of
formwork is required, the strength at an early age needs to be known.
Data on the development of strength of concretes made with water/cement ratios of
0.40, 0.53, and 0.71 are shown in Fig. 6.32 for concretes made with Type I cement in
1948 are continuously kept wet.6.117

324
 Fig. 6.32. Development of strength of concrete (determined on 150 mm (6 in.)
modified cubes) over a period of 20 years; storage under moist conditions6.117
As far as the really long-term strength is concerned, American Portland cements
made at the beginning of the century (which had a high C2S content and a low specific
surface) led to an increase in the strength of concrete stored outdoors which was
proportional to the logarithm of age up to 50 years. The 50-year strength was typically
2.4 times the 28-day strength. However, cements made since the 1930s (with a lower
C2S content and a higher specific surface) reach their peak strength between 10 and 25
years, and thereafter undergo some retrogression of strength.6.48 German Portland
cements made in 1941, when used in concrete stored outdoors, led after 30 years to a
strength 2.3 times the 28-day strength. The relative increase in strength was greater at
higher water/cement ratios. By comparison, Portland blastfurnace cement led to a 3.1-
fold increase.6.49
Maturity of concrete
The fact that the strength of concrete increases with the progress of hydration of
cement, coupled with the fact that the rate of hydration of cement increases with an
increase in temperature, leads to the proposition that strength can be expressed as a
function of the time–temperature combination. The influence of a steady temperature
on the development of strength is shown in Fig. 6.33, obtained from tests on
specimens cast, sealed and cured at the indicated temperatures.6.11 The effect of the
temperature at the time of setting, with further storage at some other temperature, is
considered on p. 362.

325
Fig. 6.33. Ratio of strength of concrete cured at different temperatures to the 28-
day strength of concrete cured at 21 °C (70 °F) (water/cement ratio = 0.50; the
specimens were cast, sealed, and cured at the indicated temperature)6.11
As the strength of concrete depends on both age and temperature, we can say that
strength is a function of ∑(time interval × temperature), and this summation is called
maturity. The temperature is reckoned from a datum found experimentally to be
between –12 and –10 °C (11 and 14 °F). This is because at temperatures below the
freezing point of water and down to about –12 °C (11 °F) concrete shows a small
increase in strength with time, but the low temperature must not be applied, of course,
until after the concrete has set and gained sufficient strength to resist damage due to
the action of frost; a ‘waiting period’ of 24 hours is usually required. Below –12 °C
(11 °F) concrete does not appear to gain strength with time.
The datum temperature generally used is –10 °C (11 °F). The appropriateness of
this value for ages up to 28 days6.101 and for temperatures in the range of 0 to 20 °C (32
to 68 °F) has been confirmed; for higher temperatures, a higher datum may be
appropriate.6.100 ASTM C 1074-04 describes a method of determination of the datum
temperature.
Maturity is measured in degree C-hours (degree F-hours) or degree C-days (degree
F-days). Figures 6.34 and 6.35 show that compressive and tensile strengths plotted
against the logarithm of maturity give a straight line.6.50 It is, therefore, possible to
express strength S2 at any maturity as a percentage of strength of concrete at any other
maturity S1; the latter is often taken as 19 800 °C h (35 600 °F h), being the maturity
of concrete cured at 18 °C (64 °F) for 28 days. This ratio of strengths, expressed as a
percentage, can then be written as:

326
 Fig. 6.34. Relation between logarithm of maturity and compressive strength of
cubes6.42

327
 Fig. 6.35. Relation between logarithm of maturity and splitting strength (tests
carried out at 2, 13, and 23 °C (35, 55, and 73 °F) up to 42 days)6.50

S1/S2 = A + B log10(maturity × 10–3).

The values of the coefficients A and B depend on the level of strength of the concrete,
that is on the water/cement ratio; those suggested by Plowman6.42 are given in Table
6.2.

Table 6.2. Plowman’s Coefficients for the Maturity Equation6.42

328
 From Fig. 6.36, it can be seen that the linearity of the relation between strength and
the logarithm of maturity applies only above a certain minimum maturity. The same
figure shows that the relation depends on the water/cement ratio; it also depends on
the type of cement used, especially if blended.

Fig. 6.36. Relation between compressive strength of ordinary Portland (Type I)

cement concrete and maturity for the data of Gruenwald6.51 as treated by Lew
and Reichard6.55
Furthermore, the early temperature also affects the precise strength–maturity
relation, including its shape.6.43 In particular, the effects of a period of exposure to a
higher temperature are not the same when this occurs immediately after casting or
later in the life of the concrete. Specifically, early high temperature leads to a lower
strength for a given total maturity than when heating is delayed for at least a week or

329
is absent. Concrete stored at 60 to 80 °C (140 to 170 °F) was found to have a long-
term strength of about 70 per cent of the strength of concrete stored at 20 °C (68 °F),
but the long-term strength was reached faster at the higher temperature.6.102 The
influence of initial temperature on the late-age strength at a given maturity was
confirmed by Carino.6.99 This is of interest in connection with steam curing. The
general topic of influence of temperature on strength is considered in Chapter 8.
The fact that the original strength–maturity relation is not applicable over a wide
range of conditions has encouraged some investigators to develop ‘improved’
maturity functions. Some of these are indeed improvements, but at the expense of
introducing complications in the development and use of the functions. Other
modified maturity functions offer improved prediction of strength in one range of
ages and temperatures but, in some other range, the prediction fits less well. One
approach uses the conversion of a curing interval at any temperature to an equivalent
interval at the reference temperature, usually 20 °C (68 °F). The concept used is that
of equivalent age, that is, the age at the reference temperature at which the same
proportion of the ultimate strength is reached as would occur at other temperatures.6.97
Despite these criticisms and developments in laboratory methods, it is reasonable
to maintain that the original maturity function, as proposed by Plowman,6.42 is a useful
tool for use in practice; ASTM Standards C 918-07 and C 1074-04 are helpful in this
respect.
ASTM C 918-07 makes the important point that there is no simple relation
between the strength of the concrete in the structure and the strength of companion
specimens, however closely they are intended to simulate the concrete in situ; only an
indication can be obtained. In view of this, ASTM C 918-07 considers that the use of
the maturity equation, developed from tests on standard compressive strength test
specimens, is as good a method of estimating the potential strength of concrete at any
required age as direct strength determination. The compressive strength specimens
must be tested at ages from not less than 24 hours and extending up to the age at
which the strength estimate is required, usually 28 days. The maturity relation is
established from a plot of strength versus the logarithm of maturity. The slope of this
line, b, makes it possible to estimate the strength S2 at maturity m2, from the
strength S1 at maturity m1, using the equation:

S2 = S1 + b(log m2 – log m1).

Clearly, the relation applies only to concrete of the given composition.

If it is intended to estimate the strength of concrete with a known temperature
history, ASTM C 1074-04 provides for the development and use of a maturity
function. This is of value when a decision has to be taken on the removal of formwork
and falsework (shoring) or on post-tensioning in prestressed concrete or on
termination of cold-weather protection.
Maturity meters are available commercially; they are clock-driven temperature
gauges, inserted into concrete, which integrate the temperature of the concrete with
respect to time and give a read-out in degree C-hours. The use of such meters removes
the uncertainty about strength in the periods of variable temperature (which can occur
accidentally even in a precast concrete factory) as the meters determine the actual
temperature of the concrete and can be located in temperature-sensitive parts of the
concrete.6.98

330
 The maturity equation should be used for wet-cured concrete only.6.44 Attempts to
allow for the relative humidity under other conditions of storage have been
made6.101 but they are unlikely to be of value as the effect of ambient relative humidity
depends on the size and shape of the concrete element.
Relation between compressive and tensile strengths
The compressive strength of concrete is its property commonly considered in
structural design but for some purposes the tensile strength is of interest; examples of
these are the design of highway and airfield slabs, shear strength, and resistance to
cracking. From the discussion of the nature of strength of concrete, it would be
expected that the two types of strength are closely related. This is indeed the case but
there is no direct proportionality, the ratio of the two strengths depending on the
general level of strength of the concrete. In other words, as the compressive
strength, fc, increases, the tensile strength, ft, also increases but at a decreasing rate.
A number of factors affect the relation between the two strengths. The beneficial
effect of crushed coarse aggregate on flexural strength was discussed on p. 287, but it
seems that the properties of fine aggregate also influence the ft/fc ratio.6.27 The ratio is
furthermore affected by the grading of the aggregate.6.28 This is probably due to the
different magnitude of the wall effect in beams and in compression specimens: their
surface/volume ratios are dissimilar so that different quantities of mortar are required
for full compaction.
Age is also a factor in the relation between ft and fc: beyond about one month, the
tensile strength increases more slowly than the compressive strength so that the
ratio ft/fc decreases with time.6.29,6.103 This is in agreement with the general tendency of
the ratio to decrease with an increase in fc.
The tensile strength of concrete can be measured by radically different tests,
namely flexure, direct tension, and splitting, and the resulting values of strength are
not the same, as discussed in Chapter 12. Consequently, the numerical value of the
ratio of the tensile strength to the compressive strength is also not the same.
Incidentally, the value of the compressive strength is also not unique but is affected
by the shape of the test specimen (see Chapter 12). For these reasons, in expressing
the ratio of the tensile to compressive strengths, the test method must be explicitly
stated. An example of the relation between the splitting strength and the compressive
strength of standard cylinders, obtained by Oluokun6.106 from a wide range of tests by
different investigators, is shown in Fig. 6.37. If the value of the flexural strength is of
interest, a factor relating the splitting strength to flexural strength needs to be
applied.6.104

331
 Fig. 6.37. Relation between splitting tensile strength and compressive strength
(measured on standard cylinders) for tests by various investigators (collated by
Oluokun)6.106
The tensile strength of concrete is more sensitive to inadequate curing than the
compressive strength,6.30 possibly because the effects of non-uniform shrinkage of
flexure test beams are very serious. Thus, air-cured concrete has a lower ft/fc ratio than
concrete cured in water and tested wet. Air entrainment affects the ft/fc ratio because
the presence of air lowers the compressive strength of concrete more than the tensile
strength, particularly in the case of rich and strong mixes.6.30 The influence of
incomplete compaction is similar to that of entrained air.6.31
Lightweight concrete conforms broadly to the pattern of the relation
between ft and fc for ordinary concrete. At very low strengths (say, 2 MPa (300 psi))
the ratio ft/fc can be as high as 0.3, but at higher strengths it is the same as for ordinary
concrete. However, drying reduces the ratio by some 20 per cent so that in the design
of lightweight concrete a reduced value of ft/fc is used.
A number of empirical formulae connecting ft and fc have been suggested, many of
them of the type:

332
 ft = k(fc)n

where k and n are coefficients. Values of n between and have been suggested.
The former value is used by the American Concrete Institute, but Gardner
and Poon6.120 found a value nearer the latter, cylinders being used in both cases.
Probably the best fit overall is given by the expression:

ft = 0.3(fc)2/3

where ft is the splitting strength, and fc is the compressive strength of cylinders, both in
megapascals. If the stress is expressed in pounds per square inch, the coefficient 0.3 is
replaced by 1.7. The above expression was suggested by Raphael.6.110 A modification
by Oluokun6.106 is:

ft = 0.2(fc)0.7

where the strengths are in megapascals; the coefficient becomes 1.4 in pounds per
square inch.
An expression used in the British Code of Practice BS 8007 : 1987 (superseded by
BS EN 1992-3 : 2006 Eurocode 2) is similar, namely:

ft = 0.12(fc)0.7

bearing in mind that the compressive strength is determined on cubes (in

megapascals); ft represents the direct tensile strength.
The differences between the various expressions are not large. What is important,
however, is that the power exponent used in the ACI Building Code 318-026.118 is too
low so that the splitting strength is overestimated at low compressive strengths and
underestimated at high compressive strengths.6.105
Bond between concrete and reinforcement
Since structural concrete is, in the vast majority of cases, used with steel
reinforcement, the strength of bond between the two materials is of considerable
importance with respect to structural behaviour, including cracking due to shrinkage
and early thermal effects. Bond arises primarily from friction and adhesion between
concrete and steel, and from mechanical interlocking in the case of deformed bars.
Bond may also be beneficially affected by the shrinkage of concrete relative to the
steel.
In a structure, the bond strength involves not only the properties of the concrete but
also other factors. These include the geometry of the reinforcement and of the
structure such as the thickness of cover to the reinforcement. The state of the surface
of the steel is also a factor. The presence of rust on the surface of the steel, provided
the rust is well connected to the underlying steel, improves bond of plain bars and
does not impair the bond of deformed reinforcement.6.108 Coating by galvanizing or by
epoxy affects the bond strength.
These considerations put the subject of bond largely outside the scope of the
present book except in so far as the properties of concrete influence the bond strength
which, incidentally, is not easily determined.

333
 The critical property is the tensile strength of concrete. For this reason, design
formulae for bond strength usually express it as being proportional to the square root
of compressive strength. As shown earlier, the tensile strength of concrete is
proportional to a somewhat higher power of the compressive strength, say about 0.7;
consequently, the expressions used in the various codes are not a correct
representation of the indirect dependence of the bond strength on the compressive
strength of concrete. Nevertheless, bond strength of deformed steel bars has been
shown to increase with an increase in compressive strength, albeit at a decreasing rate,
for concrete strengths up to about 95 MPa (14 000 psi).6.107,6.109
A rise in temperature reduces the bond strength of concrete: at 200 to 300 °C (400
to 570 °F) there may be a loss of one-half of the bond strength at room temperature.
References
6.1. K. K. SCHILLER, Porosity and strength of brittle solids (with particular
reference to gypsum), Mechanical Properties of Non-metallic Brittle
Materials, pp. 35–45 (London, Butterworth, 1958).
6.2. NATIONAL SAND AND GRAVEL ASSOCIATION, Joint Tech. Information Letter No.
155 (Washington DC, 29 April 1959).
6.3. A. HUMMEL, Das Beton – ABC (Berlin, W. Ernst, 1959).
6.4. A. M. NEVILLE, Tests on the strength of high-alumina cement concrete, J.
New Zealand Inst. E., 14, No. 3, pp. 73–7 (1959).
6.5. T. C. POWERS, The non-evaporable water content of hardened portland cement
paste: its significance for concrete research and its method of
determination, ASTM Bull. No. 158, pp. 68–76 (May 1949).
6.6. T. C. POWERS and T. L. BROWNYARD, Studies of the physical properties of
hardened portland cement paste (Nine parts), J. Amer. Concr. Inst., 43 (Oct.
1946 to April 1947).
6.7. T. C. POWERS, The physical structure and engineering properties of
concrete, Portl. Cem. Assoc. Res. Dept. Bull. 90 (Chicago, July 1958).
6.8. T. C. POWERS, Structure and physical properties of hardened portland cement
paste, J. Amer. Ceramic Soc., 41, pp. 1–6 (Jan. 1958).
6.9. L. J. M. NELISSEN, Biaxial testing of normal concrete, Heron, 18, No. 1, pp.
1–90 (1972).
6.10. M. A. WARD, A. M. NEVILLE and S. P. SINGH, Creep of air-entrained
concrete, Mag. Concr. Res., 21, No. 69, pp. 205–10 (1969).
6.11. W. H. PRICE, Factors influencing concrete strength, J. Amer. Concr. Inst., 47,
pp. 417–32 (Feb. 1951).
6.12. H. C. ERNTROY and B. W. SHACKLOCK, Design of high-strength concrete
mixes, Proc. of a Symposium on Mix Design and Quality Control of
Concrete, pp. 55–73 (Cement and Concrete Assoc., London, May 1954).
6.13. B. G. SINGH, Specific surface of aggregates related to compressive and
flexural strength of concrete, J. Amer. Concr. Inst., 54, pp. 897–907 (April
1958).
6.14. A. M. NEVILLE, Some aspects of the strength of concrete, Civil
Engineering (London), 54, Part 1, Oct. 1959, pp. 1153–6; Part 2, Nov. 1959,
pp. 1308–10; Part 3, Dec. 1959, pp. 1435–8.

334
 6.15. A. M. NEVILLE, The influence of the direction of loading on the strength of
concrete test cubes, ASTM Bull. No. 239, pp. 63–5 (July 1959).
6.16. E. OROWAN, Fracture and strength of solids, Reports on Progress in
Physics, 12, pp. 185–232 (Physical Society, London, 1948–49).
6.17. A. A. GRIFFITH, The phenomena of rupture and flow in solids, Philosophical
Transactions, Series A, 221, pp. 163–98 (Royal Society, 1920).
6.18. A. M. NEVILLE, The failure of concrete compression test specimens, Civil
Engineering (London), 52, pp. 773–4 (July 1957).
6.19. R. JONES and M. F. KAPLAN, The effects of coarse aggregate on the mode of
failure of concrete in compression and flexure, Mag. Concr. Res., 9, No. 26,
pp. 89–94 (1957).
6.20. F. M. LEA, Cement research: retrospect and prospect, Proc. 4th Int. Symp. on
the Chemistry of Cement, pp. 5–8 (Washington DC, 1960).
6.21. O. Y. BERG, Strength and plasticity of concrete, Doklady Akademii Nauk
S.S.S.R., 70, No. 4, pp. 617–20 (1950).
6.22. R. L’HERMITE, Idées actuelles sur la technologie du béton, Institut Technique
du Bâtiment et des Travaux Publics (Paris, 1955).
6.23. R. JONES and E. N. GATFIELD, Testing concrete by an ultrasonic pulse
technique, Road Research Tech. Paper No. 34 (HMSO, London, 1955).
6.24. W. KUCZYNSKI, Wplyw kruszywa grubego na wytrzymalosć betonu
(L’influence de l’emploi d’agrégats gros sur la résistance du
béton). Archiwum Inzynierii Ladowej, 4, No. 2, pp. 181–209 (1958).
6.25. M. F. KAPLAN, Flexural and compressive strength of concrete as affected by
the properties of coarse aggregates, J. Amer. Concr. Inst., 55, pp. 1193–208
(May 1959).
6.26. US BUREAU OF RECLAMATION, Triaxial strength tests of neat cement and mortar
cylinders, Concrete Laboratory Report No. C-779 (Denver, Colorado, Nov.
1954).
6.27. P. J. F. WRIGHT, Crushing and flexural strengths of concrete made with
limestone aggregate, Road Res. Lab. Note RN/3320/PJFW (London, HMSO,
Oct. 1958).
6.28. L. SHUMAN and J. TUCKER, J. Res. Nat. Bur. Stand. Paper No. RP1552, 31, pp.
107–24 (1943).
6.29. A. G. A. SAUL, A comparison of the compressive, flexural, and tensile
strengths of concrete, Cement Concr. Assoc. Tech. Rep. TRA/333 (London,
June 1960).
6.30. B. W. SHACKLOCK and P. W. KEENE, Comparison of the compressive and
flexural strengths of concrete with and without entrained air, Civil
Engineering (London), 54, pp. 77–80 (Jan. 1959).
6.31. M. F. KAPLAN, Effects of incomplete consolidation on compressive and
flexural strength, ultrasonic pulse velocity, and dynamic modulus of
elasticity of concrete, J. Amer. Concr. Inst., 56, pp. 853–67 (March 1960).
6.32. G. VERBECK, Energetics of the hydration of Portland cement, Proc. 4th Int.
Symp. on the Chemistry of Cement pp. 453–65, Washington DC (1960).

335
 6.33. A. GRUDEMO, Development of strength properties of hydrating cement pastes
and their relation to structural features, Proc. Symp. on Some Recent
Research on Cement Hydration, 8 pp. (CEMBUREAU, 1975).
6.34. D. M. ROY and G. R. GOUDA, Porosity–strength relation in cementitious
materials with very high strengths, J. Amer. Ceramic Soc., 53, No. 10, pp.
549–50 (1973).
6.35. R. F. FELDMAN and J. J. BEAUDOIN, Microstructure and strength of hydrated
cement, Cement and Concrete Research, 6, No. 3, pp. 389–400 (1976).
6.36. P. G. LOWE, Deformation and fracture of plain concrete, Mag. Concr.
Res., 30, No. 105, pp. 200–4 (1978).
6.37. S. D. SANTIAGO and H. K. HILSDORF, Fracture mechanisms of concrete under
compressive loads, Cement and Concrete Research, 3, No. 4, pp. 363–88
(1973).
6.38. C. PERRY and J. E. GILLOTT, The influence of mortar–aggregate bond strength
on the behaviour of concrete in uniaxial compression, Cement and Concrete
Research, 7, No. 5, pp. 553–64 (1977).
6.39. R. E. FRANKLIN and T. M. J. KING, Relations between compressive and
indirect-tensile strengths of concrete, Road. Res. Lab. Rep. LR412, 32 pp.
(Crowthorne, Berks., 1971).
6.40. A. F. STOCK, D. J. HANNANT and R. I. T. WILLIAMS, The effect of aggregate
concentration upon the strength and modulus of elasticity of concrete, Mag.
Concr. Res., 31, No. 109, pp. 225–34 (1979).
6.41. H. KAWAKAMI, Effect of gravel size on strength of concrete with particular
reference to sand content, Proc. Int. Conf. on Mechanical Behaviour of
Materials, Kyoto, 1971 – Vol. IV, Concrete and Cement Paste Glass and
Ceramics, pp. 96–103 (Kyoto, Japan, Society of Materials Science, 1972).
6.42. J. M. PLOWMAN, Maturity and the strength of concrete, Mag. Concr. Res., 8,
No. 22, pp. 13–22 (1956).
6.43. P. KLIEGER, Effect of mixing and curing temperature on concrete strength, J.
Amer. Concr. Inst., 54, pp. 1063–81 (June 1958).
6.44. P. KLIEGER, Discussion on: Maturity and the strength of concrete, Mag. Concr.
Res., 8, No. 24, pp. 175–8 (1956).
6.45. C. D. POMEROY, D. C. SPOONER and D. W. HOBBS, The dependence of the
compressive strength of concrete on aggregate volume concentration for
different shapes of specimen, Cement Concr. Assoc. Departmental Note DN
4016, 17 pp. (Slough, U.K., March 1971).
6.46. D. W. HOBBS, C. D. POMEROY and J. B. NEWMAN, Design stresses for concrete
structures subject to multiaxial stresses, The Structural Engineer., 55, No. 4,
pp. 151–64 (1977).
6.47. D. W. HOBBS, Strength and deformation properties of plain concrete subject
to combined stress, Part 3: results obtained on a range of flint gravel
aggregate concretes, Cement Concr. Assoc. Tech. Rep. TRA/42.497, 20 pp.
(London, July 1974).
6.48. G. W. WASHA and K. F. WENDT, Fifty year properties of concrete, J. Amer.
Concr. Inst., 72, No. 1, pp. 20–8 (1975).

336
 6.49. K. WALZ, Festigkeitsentwicklung von Beton bis zum Alter von 30 und 50
Jahren, Beton, 26, No. 3, pp. 95–8 (1976).
6.50. H. S. LEW and T. W. REICHARD, Mechanical properties of concrete at early
ages, J. Amer. Concr. Inst., 75, No. 10, pp. 533–42 (1978).
6.51. E. GRUENWALD, Cold weather concreting with high-early strength
cement, Proc. RILEM Symp. on Winter Concreting, Theory and Practice,
Copenhagen, 1956, 30 pp. (Danish National Inst., of Building Research,
1956).
6.52. K. NEWMAN, Criteria for the behaviour of plain concrete under complex states
of stress, Proc. Int. Conf. on the Structure of Concrete, London, Sept. 1965,
pp. 255–74 (London, Cement and Concrete Assoc., 1968).
6.53. N. J. CARINO and F. O. SLATE, Limiting tensile strain criterion for failure of
concrete, J. Amer. Concr. Inst., 73, No. 3, pp. 160–5 (1976).
6.54. M. E. TASUJI, A. H. NILSON and F. O. SLATE, Biaxial stress–strain relationships
for concrete, Mag. Concr. Res., 31, No. 109, pp. 217–24 (1979).
6.55. H. S. LEW and T. W. REICHARD, Prediction of strength of concrete from
maturity, in Accelerated Strength Testing, ACI SP-56, pp. 229–48 (Detroit,
Michigan, 1978).
6.56. O. Y. BERG, Research on the concrete strength theory, Building Research and
Documentation, Contributions and Discussions, First CIB Congress, pp. 60–
9 (Rotterdam, 1959).
6.57. L. A. LARBI, Microstructure of the interfacial zone around aggregate particles
in concrete, Heron, 38, No. 1, 69 pp. (1993).
6.58. M. KAKIZAKI, H. EDAHIRO, T. TOCHIGI and T. NIKI, Effect of Mixing Method on
Mechanical Properties and Pore Structure of Ultra High-Strength Concrete,
Katri Report No. 90, 19 pp. (Kajima Corporation, Tokyo, 1992) (and also in
ACI SP-132 (Detroit, Michigan, 1992)).
6.59. L. F. NIELSEN, Strength development in hardened cement paste: examination
of some empirical equations, Materials and Structures, 26, No. 159, pp.
255–60 (1993).
6.60. S. WALKER and D. L. BLOEM, Studies of flexural strength of concrete, Part 3:
Effects of variation in testing procedures, Proc. ASTM, 57, pp. 1122–39
(1957).
6.61. H. GREEN, Impact testing of concrete, Mechanical Properties of Non-metallic
Brittle Materials, pp. 300–13 (London, Butterworth, 1958).
6.62. B. MATHER, Comment on “Water-cement ratio is passé”, Concrete
International, 11, No. 11, p. 77 (1989).
6.63. M. RÖSSLER and I. ODLER, Investigations on the relationship between porosity,
structure and strength of hydrated Portland cement pastes. I. Effect of
porosity, Cement and Concrete Research, 15, No. 2, pp. 320–30 (1985).
6.64. I. ODLER and M. RÖSSLER, Investigations on the relationship between porosity,
structure and strength of hydrated Portland cement pastes. II. Effect of pore
structure and the degree of hydration, Cement and Concrete Research, 15,
No. 3, pp. 401–10 (1985).

337
 6.65. J. J. BEAUDOIN and V. S. RAMACHANDRAN, A new perspective on the hydration
characteristics of cement phases, Cement and Concrete Research, 22, No. 4,
pp. 689–94 (1992).
6.66. R. SERSALE, R. CIOFFI, G. FRIGIONE and F. ZENONE, Relationship between
gypsum content, porosity, and strength of cement, Cement and Concrete
Research, 21, No. 1, pp. 120–6 (1991).
6.67. R. F. FELDMAN, Application of the helium inflow technique for measuring
surface area and hydraulic radius of hydrated Portland cement, Cement and
Concrete Research, 10, No. 5, pp. 657–64 (1980).
6.68. D. WINSLOW and DING LIU, The pore structure of paste in concrete, Cement
and Concrete Research, 20, No. 2, pp. 227–84 (1990).
6.69. R. L. DAY and B. K. MARSH, Measurement of porosity in blended cement
pastes, Cement and Concrete Research, 18, No. 1, pp. 63–73 (1988).
6.70. A. A. RAHMAN, Characterization of the porosity of hydrated cement pastes,
in The Chemistry and Chemically-Related Properties of Concrete, Ed. F. P.
Glasser, British Ceramic Proceedings No. 35, pp. 249–63 (Stoke-on-Trent,
1984).
6.71. L. I. KNAB, J. R. CLIFTON and J. B. INGE, Effects of maximum void size and
aggregate characteristics on the strength of mortar, Cement and Concrete
Research, 13, No. 3, pp. 383–90 (1983).
6.72. E. M. KROKOSKY, Strength vs. structure: a study for hydraulic
cements, Materials and Structures, 3, No. 17, pp. 313–23 (Paris, Sept.–Oct.
1970).
6.73. E. RYSHKEWICH, Compression strength of porous sintered alumina and
zirconia, J. Amer. Ceramic Soc., 36, pp. 66–8 (Feb. 1953).
6.74. Discussion of paper by H. J. GILKEY: Water/cement ratio versus strength –
another look, J. Amer. Concr. Inst., Part 2, 58, pp. 1851–78 (Dec. 1961).
6.75. D. W. HOBBS, Strength and deformation properties of plain concrete subject
to combined stress, Part 1: strength results obtained on one concrete, Cement
Concr. Assoc. Tech. Rep. TRA/42.451 (London, Nov. 1970).
6.76. T. T. C. Hsu, F. O. SLATE, G. M. STURMAN and G. WINTER, Microcracking of
plain concrete and the shape of the stress–strain curve, J. Amer. Concr.
Inst., 60, pp. 209–24 (Feb. 1963).
6.77. B. MATHER, What do we need to know about the response of plain concrete
and its matrix to combined loadings?, Proc. 1st Conf. on the Behavior of
Structural Concrete Subjected to Combined Loadings, pp. 7–9 (West
Virginia Univ., 1969).
6.78. H. KUPFER, H. K. HILSDORF and H. RÜSCH, Behaviour of concrete under biaxial
stresses, J. Amer. Concr. Inst., 66, pp. 656–66 (Aug. 1969).
6.79. B. BRESLER and K. S. PISTER, Strength of concrete under combined stresses, J.
Amer. Concr. Inst., 55, pp. 321–45 (Sept. 1958).
6.80. S. POPOVICS, Analysis of the concrete strength versus water–cement ratio
relationship, ACI Materials Journal, 57, No. 5, pp. 517–29 (1990).

338
 6.81. N. McN. ALFORD, G. W. GROVES and D. D. DOUBLE, Physical properties of high
strength cement paste, Cement and Concrete Research, 12, No. 3, pp. 349–
58 (1982).
6.82. Z. P. BAžANT, F. C. BISHOP and TA-PENG CHANG, Confined compression tests of
cement paste and concrete up to 300 ksi, ACI Journal, 83, No. 4, pp. 553–60
(1986).
6.83. A. MEYER, Über den Einfluss des Wasserzementwertes auf die Frühfestigkeit
von Beton, Betonstein Zeitung, No. 8, pp. 391–4 (1963).
6.84. L. BJERKELI, J. J. JENSEN and R. LENSCHOW, Strain development and static
compressive strength of concrete exposed to water pressure loading, ACI
Structural Journal, 90, No. 3, pp. 310–15 (1993).
6.85. CHUAN-ZHI WANG, ZHEN-HAI GUO and XIU-QIN, ZHANG, Experimental
investigation of biaxial and triaxial compressive concrete strength, ACI
Materials Journal, 84, No. 2, pp. 92–6 (1987).
6.86. W. S. YIN, E. C. M. SU, M. A. MANSUR and T. C. Hsu, Biaxial tests of plain
and fiber concrete, ACI Materials Journal, 86, No. 3, pp. 236–43 (1989).
6.87. S. P. SHAH, Fracture toughness for high-strength concrete, ACI Materials
Journal, 87, No. 3, pp. 260–5 (1990).
6.88. G. GIACCIO, C. Rocco and R. ZERBINO, The fracture energy (GF) of high-
strength concretes, Materials and Structures, 26, No. 161, pp. 381–6 (1993).
6.89. MUN-HONG ZHANG and O. E. GJØRV, Microstructure of the interfacial zone
between lightweight aggregate and cement paste, Cement and Concrete
Research, 20, No. 4, pp. 610–18 (1990).
6.90. M. M. SMADI and F. O. SLATE, Microcracking of high and normal strength
concretes under short- and long-term loadings, ACI Materials Journal, 86,
No. 2, pp. 117–27 (1989).
6.91. F. O. SLATE and K. C. HOVER, Microcracking in concrete, in Fracture
Mechanics of Concrete: Material Characterization and Testing, Eds A.
Carpinteri and A. R. Ingraffea, pp. 137–58 (The Hague, Martinus Nijhoff,
1984).
6.92. A. JORNET, E. GUIDALI and U. MÜHLETHALER, Microcracking in high
performance concrete, in Proceedings of the Fourth Euroseminar on
Microscopy Applied to Building Materials, Eds J. E. Lindqvist and B. Nitz,
Sp. Report 1993: 15, 6 pp. (Swedish National Testing and Research Institute:
Building Technology, 1993).
6.93. P. J. M. MONTEIRO, J. C. MASO and J. P. OLLIVIER, The aggregate–mortar
interface, Cement and Concrete Research, 15, No. 6, pp. 953–8 (1985).
6.94. K. L. SCRIVENER and E. M. GARINER, Microstructural gradients in cement paste
around aggregate particles, Materials Research Symposium Proc., 114, pp.
77–85 (1988).
6.95. XIE PING, J. J. BEAUDOIN and R. BROUSSEAU, Effect of aggregate size on the
transition zone properties at the Portland cement paste interface, Cement and
Concrete Research, 21, No. 6, pp. 999–1005 (1991).
6.96. J. C. MASO, La liaison pâte-granulats, in Le Béton Hydraulique, Eds J. Baron
and R. Sauterey, pp. 247–59 (Presses de l’École Nationale des Ponts et
Chaussées, Paris, 1982).

339
 6.97. N. J. CARINO and R. C. TANK, Maturity functions for concretes made with
various cements and admixtures, ACI Materials Journal, 89, No. 2, pp. 188–
96 (1992).
6.98. R. I. PEARSON, Maturity meter speeds post-tensioning of structural concrete
frame, Concrete International, 9, No. 5, pp. 63–4 (April 1987).
6.99. N. J. CARINO and H. S. LEW, Temperature effects on strength–maturity
relations of mortar, ACI Journal, 80, No. 3, pp. 177–82 (1983).
6.100. N. J. CARINO, The maturity method: theory and
application, Cement, Concrete, and Aggregates, 6, No. 2, pp. 61–73 (1984).
6.101. K. AYUTA, M. HAYASHI and H. SAKURAI, Relation between concrete strength
and cumulative temperature, Cement Association of Japan Review, pp. 236–
9 (1988).
6.102. E. GAUTHIER and M. REGOURD, The hardening of cement in function of
temperature, in Proceedings of RILEM International Conference on
Concrete of Early Ages, Vol. 1, pp. 145–55 (Paris, Anciens ENPC, 1982).
6.103. K. KOMLOS, Comments on the long-term tensile strength of plain
concrete, Mag. Concr. Res., 22, No. 73, pp. 232–8 (1970).
6.104. L. BORTOLOTTI, Interdependence of concrete strength parameters, ACI
Materials Journal, 87, No. 1, pp. 25–6 (1990).
6.105. N. J. CARINO and H. S. LEW, Re-examination of the relation between splitting
tensile and compressive strength of normal weight concrete, ACI Journal, 79,
No. 3, pp. 214–19 (1982).
6.106. F. A. OLUOKUN, Prediction of concrete tensile strength from compressive
strength: evaluation of existing relations for normal weight concrete, ACI
Materials Journal, 88, No. 3, pp. 302–9 (1991).
6.107. O. E. GJØRV, P. J. M. MONTEIRO and P. K. MEHTA, Effect of condensed silica
fume on the steel–concrete bond, ACI Materials Journal, 87, No. 6, pp. 573–
80 (1990).
6.108. F. G. MURPHY, The Effect of Initial Rusting on the Bond Performance of
Reinforcement, CIRIA Report 71, 36 pp. (London, 1977).
6.109. I. SCHALLER, F. DE LARRARD and J. FUCHS, Adhérence des armatures passives
dans le béton à très hautes performances, Bulletin liaison Labo. Ponts et
Chaussées, 167, pp. 13–21 (May–June 1990).
6.110. J. M. RAPHAEL, Tensile strength of concrete, ACI Materials Journal, 81, No. 2,
pp. 158–65 (1984).
6.111. E. K. ATTIOGBE and D. DARWIN, Submicrocracking in cement paste and
mortar, ACI Materials Journal, 84, No. 6, pp. 491–500 (1987).
6.112. K. M. ALEXANDER and I. IVANUSEC, Long term effects of cement SO3 content on
the properties of normal and high-strength concrete, Part I. The effect on
strength, Cement and Concrete Research, 12, No. 1, pp. 51–60 (1982).
6.113. N. HEARN and R. D. HOOTON, Sample mass and dimension effects on mercury
intrusion porosimetry results, Cement and Concrete Research, 22, No. 5, pp.
970–80 (1992).

340
 6.114. R. A. COOK and K. C. HOVER, Mercury porosimetry of cement-based materials
and associated correction factors, ACI Materials Journal, 90, No. 2, pp.
152–61 (1993).
6.115. N. HEARN, R. D. HOOTON and R. H. MILLS, Pore structure and permeability,
in Concrete and Concrete-Making Materials, ASTM Sp. Tech. Publ. No.
169C pp. 241–62 (Philadelphia, 1994).
6.116. W. S. NAJJAR and K. C. HOVER, Neutron radiography for microcrack studies of
concrete cylinders subjected to concentric and excentric compressive
loads, ACI Materials Journal, 86, No. 4, pp. 354–9 (1989).
6.117. S. L. WOOD, Evaluation of the long-term properties of concrete, ACI
Materials Journal, 88, No. 6, pp. 630–43 (1991).
6.118. ACI 318-02, Building code requirements for structural concrete, ACI
Manual of Concrete Practice, Part 3: Use of Concrete in Buildings – Design,
Specifications, and Related Topics, 443 pp.
6.119. D. C. SPOONER, C. D. POMEROY and J. W. DOUGILL, Damage and energy
dissipation in cement pastes in compression, Mag. Concr. Res., 28, No. 94,
pp. 21–9 (1976).
6.120. N. J. GARDNER and S. M. POON, Time and temperature effects on tensile, bond,
and compressive strengths, J. Amer. Concr. Inst., 73, No. 7, pp. 405–9
(1976).

341
Chapter 7. Further aspects of hardened concrete

In the preceding chapter, we considered the main factors influencing the strength of
concrete. Here, some further aspects of strength will be discussed, including fatigue
and impact; this will be followed by a brief description of electrical and acoustic
properties of concrete.
Curing of concrete
In order to obtain good concrete, the placing of an appropriate mix must be followed
by curing in a suitable environment during the early stages of hardening. Curing is the
name given to procedures used for promoting the hydration of cement, and consists of
a control of temperature and of the moisture movement from and into the concrete.
The temperature factor is dealt with in Chapter 8.
More specifically, the object of curing is to keep concrete saturated, or as nearly
saturated as possible, until the originally water-filled space in the fresh cement paste
has been filled to the desired extent by the products of hydration of cement. In the
case of site concrete, active curing stops nearly always long before the maximum
possible hydration has taken place.
Powers7.36 showed that hydration is greatly reduced when the relative humidity
within the capillary pores drops below 80 per cent; this was confirmed by Patel et
al.7.3 Hydration at a maximum rate can proceed only under conditions of
saturation. Figure 7.1 shows the degree of hydration of cement after six months’
storage at different relative humidities, and it is clear that, below a vapour pressure of
0.8 of the saturation pressure, the degree of hydration is low, and is negligible below
0.3 of the saturation pressure.7.36

342
 Fig. 7.1. Water taken up by dry cement exposed for six months to different
vapour pressures7.36
It follows that, for hydration to continue, the relative humidity inside the concrete
has to be maintained at a minimum of 80 per cent. If the relative humidity of the
ambient air is at least that high, there will be little movement of water between the
concrete and the ambient air, and no active curing is needed to ensure continuing
hydration. Strictly speaking, the preceding statement is valid only if no other factors
intervene, e.g. there is no wind, there is no difference in temperature between the
concrete and the air, and if the concrete is not exposed to solar radiation. In practice,

343
therefore, active curing is unnecessary only in a very humid climate with a steady
temperature. It is important to note that in many parts of the world the relative
humidity falls below 80 per cent at some time during the day so that the belief in
‘natural curing’, just because the weather is wet, is unfounded.
An indication of the influence on evaporation from the concrete surface, of
temperature and relative humidity of the surrounding air, and of the wind velocity is
given in Figs 7.2, 7.3, and 7.4, based on Lerch’s7.37 results. The difference between the
temperatures of concrete and of air also affects the loss of water, as shown in Fig. 7.5.
Thus, concrete saturated in day-time would lose water during a cold night, and this
would also be the case with concrete cast in cold weather, even in saturated air. The
examples quoted are merely typical as the actual loss of water depends on the
surface/volume ratio of the specimen.7.38

Fig. 7.2. Influence of relative humidity of air on the loss of water from concrete
in the early stages after placing (air temperature 21 °C (70 °F); wind velocity 4.5
m/s (10 mph))

344
 Fig. 7.3. Influence of temperature of air and concrete on the loss of water from
concrete in the early stages after placing (relative humidity of air 70 per cent;
wind velocity 4.5 m/s (10 mph))

345
Fig. 7.4. Influence of wind velocity on the loss of water from concrete in the early
stages after placing (relative humidity of air 70 per cent, temperature 21 °C
(70 °F))

346
 Fig. 7.5. Influence of temperature of concrete (at an air temperature of 4.5 °C
(40 °F)) on the loss of water from concrete in the early stages after placing
(relative humidity of air 100 per cent, wind velocity 4.5 m/s (10 mph))
Prevention of the loss of water from the concrete is of importance not only because
the loss adversely affects the development of strength, but also because it leads to
plastic shrinkage, increased permeability, and reduced resistance to abrasion.
From the preceding discussion, it could be inferred that, for hydration of cement to
continue, it is sufficient to prevent the loss of moisture from the concrete. This is true
only if the water/cement ratio of the concrete is sufficiently high for the quantity of
the mix water to be adequate for hydration to continue. It was shown in Chapter 1 that
hydration of cement can take place only in water-filled capillaries. This is why loss of
water by evaporation from the capillaries must be prevented. Furthermore, water lost
internally by self-desiccation (due to the chemical reactions of hydration of cement)
has to be replaced by water from outside, i.e. ingress of water into the concrete must
be made possible.
It may be recalled that hydration of a sealed specimen can proceed only if the
amount of water present in the paste is at least twice that of the water already
combined. Self-desiccation is thus of importance in mixes with water/cement ratios
below about 0.5; for higher water/cement ratios, the rate of hydration of a sealed
specimen equals that of a saturated specimen.7.35 It should not be forgotten, however,
that only half the water present in the paste can be used for chemical combination;

347
this is so even if the total amount of water present is less than the water required for
chemical combination.7.36
In view of the above, we can distinguish between curing needs in situations where,
on the one hand, only loss of water from the concrete needs to be prevented and, on
the other, situations where water ingress from outside is necessary for hydration to
continue. The dividing line is approximately at a water/cement ratio of 0.5. With
many modern concretes having a water/cement ratio below 0.5, the promotion of
hydration by the ingress of water into concrete is desirable.
It should be added that concrete remote from the surface, that is at depth, is hardly
subjected to moisture movement, which affects only an outer zone, typically 30 mm
deep, but occasionally up to a depth of 50 mm. In reinforced concrete, this depth
represents all or most of the depth of cover.
Thus, concrete in the interior of a structural member is generally unaffected by
curing, so that curing is of little importance with respect to structural strength except
in the case of very thin members. On the other hand, the properties of concrete in the
outer zone are greatly influenced by curing; it is the concrete in this zone that is
subject to weathering, carbonation, and abrasion, and the permeability of the outer-
zone concrete has a paramount influence on the protection of steel reinforcement from
corrosion (see Chapter 11).
An indication of the depth of the outer zone which is affected by curing can be
obtained from Parrott’s tests7.2 on concrete with a water/cement ratio of 0.59 stored at
20 °C (68 °F) in air with a relative humidity of 60 per cent; he found the following
periods for the relative humidity inside concrete to fall to 90 per cent: 12 days to a
depth of 7.5 mm; 45 days to a depth of 15.5 mm; and 172 days to a depth of 33.5 mm.
At lower water/cement ratios, which are common in modern concrete, these periods
would be longer.
A reduction in the ambient relative humidity from 100 to 94 per cent was found
greatly to increase the water absorption capacity of the concrete, this being an
indication of the extent of the continuous large pore system in the concrete.7.5 Curing
at an external relative humidity below about 80 per cent was shown to result in a very
large increase in the volume of pores larger than 37 nm, which are relevant to the
durability of concrete.7.3
It follows from the preceding discussion that the effects of curing should be
studied on the outer-zone concrete. However, traditionally, they are expressed in
terms of the influence of curing on strength, that is on a comparison of the strength of
the specimens stored in water (or in fog) with the strength of those stored under some
other conditions for different periods; this is taken to demonstrate the effectiveness of
curing and its beneficial effect. An example of this is shown in Fig. 7.6, obtained for
concrete with a water/cement ratio of 0.50. The loss of strength due to inadequate
curing is more pronounced in smaller specimens, but the loss is smaller in lightweight
aggregate concrete.7.55 Tensile and compressive strengths are affected in a similar
manner; in both cases, richer mixes are slightly more susceptible.7.56

348
 Fig. 7.6. Influence of moist curing on the strength of concrete with a
water/cement ratio of 0.507.11
The loss of strength at 28 days seems to be directly related to the loss of water
which occurred during the first 3 days; the temperature (20 or 40 °C (68 or 104 °F))
has no effect7.7 (see Fig. 7.7).

349
 Fig. 7.7. Relation between the compressive strength of concrete at the age of 28
days and the loss of water (by mass of concrete) during the first 3 days (based on
ref. 7.7)
The effect of inadequate curing on strength is greater at higher water/cement ratios
and is also greater in concretes with a lower rate of development of strength.7.29 Thus,
the strength of concretes made with ordinary Portland (Type I) cement is more
affected by poor curing. Likewise, concretes containing fly ash or ground granulated
blastfurnace slag are more affected than concretes made with Portland cement only.
It must be stressed that, for a satisfactory development of strength, it is not
necessary for all the cement to hydrate and, indeed, this is only rarely achieved in
practice: as shown in Chapter 6, the quality of concrete depends primarily on the
gel/space ratio of the paste. If, however, the water-filled space in fresh concrete is
greater than the volume that can be filled by the products of hydration, greater
hydration will lead to a higher strength and a lower permeability.
Methods of curing
There are two broad categories of curing whose principles will now be considered,
recognizing that the actual procedures used vary widely, depending on the conditions
on site and on the size, shape, and position of the concrete member. The methods may
be broadly described as wet curing and membrane curing, respectively.
The first method is that of providing water which can be imbibed by the concrete.
This requires that the surface of the concrete is continuously in contact with water for
a specified length of time, starting as soon as the surface of the concrete is no longer
liable to damage. Such conditions can be achieved by continuous spraying or flooding
(ponding), or by covering the concrete with wet sand or earth, sawdust or straw. Some
care is required as staining may result. Periodically-wetted clean hessian (burlap) or
cotton mats (thick and lapped) may be used, or alternatively an absorbent covering
with access to water may be placed over the concrete. On inclined or vertical surfaces,
soaking hoses can be used. A continuous supply of water is naturally more efficient
than an intermittent one, and Fig. 7.8 compares the strength development of concrete
cylinders whose top surface was flooded during the first 24 hours with that of
cylinders covered with wet hessian.7.77 The difference is apparent only at water/cement
ratios below about 0.4 where self-desiccation results in a shortage of water within the
concrete. It follows that for low water/cement ratios wet curing is highly desirable.
Length of curing is dealt with in BS EN 13670-2009.

350
 Fig. 7.8. Influence of curing conditions on strength of test cylinders7.77
As far as quality of the water used for curing is concerned, ideally it should be the
same as mixing water (see p. 184). Sea water may lead to corrosion of reinforcement.
Also, iron or organic matter may cause staining, particularly if water flows slowly
over the concrete and evaporates rapidly. In some cases, discoloration is of no
significance.
Whether or not staining will take place cannot be stated on the basis of a chemical
analysis and should be checked by a performance test. U.S. Army Corps of
Engineers7.40 recommends a preliminary test in which 300 ml of the water to be used
for curing is evaporated from a slight depression, 100 mm (4 in.) in diameter, in the
surface of a specimen of neat white cement or plaster of Paris. If the resulting
colouring is not considered objectionable, a further test is performed. Here, 150 litres
(40 U.S. gallons) of water are allowed to flow lengthwise over a 150 by 150 by 750
mm (6 by 6 by 30 in.) concrete beam with a channel-shaped top surface, placed at 15
to 20° to the horizontal; the rate of flow is 4 litres in 3 to 4 hours. Forced circulation
of air and heating by electric lamps encourage evaporation and thus deposition of the
residue. The test is again evaluated by observation only and, if necessary, an actual
field test may be performed: a 2 m2 (or 20 ft2) slab is cured.

351
 It is essential that curing water be free from substances that attack hardened
concrete; these are discussed in Chapters 10 and 11.
The temperature of the water should not be much lower than that of the concrete in
order to avoid thermal shock or steep temperature gradients; ACI 308-92 recommends
a maximum difference of 11 °C (20 °F).7.9
The second method of curing relies on the prevention of loss of water from the
surface of the concrete, without the possibility of external water ingressing into it.
This could be called a water-barrier method. The techniques used include covering the
surface of the concrete with overlapping polyethylene sheeting, laid flat, or with
reinforced paper. The sheeting can be black, which is preferable in cold weather, or
white, which has the advantage of reflection of solar radiation in hot weather. Paper
with a white surface is also available. Sheeting can cause discoloration or mottling
because of non-uniform condensation of water on the underside.
Another technique uses spray-applied curing compounds which form a membrane.
The common ones are solutions of synthetic hydrocarbon resins in high-volatility
solvents, sometimes including a fugitive bright-colour dye. The dye makes obvious
the areas not properly sprayed. A white or alumina pigment can be included to reduce
the solar heat gain; this is very effective. Other resin solutions are available: acrylic,
vinyl or styrene butadiene, and chlorinated rubber. Wax emulsions can also be used,
but they result in a slippery finish which is not easy to remove, whereas the
hydrocarbon resins have poor adhesion to concrete and are degraded by ultraviolet
light; both these features are desirable.
A specification for liquid membrane-forming curing compounds is given in ASTM
C 309-07; and for sheet materials in ASTM C 171-07.
A question often arises: which curing method or technique to use? For concrete
with a water/cement ratio lower than about 0.5, and certainly lower than 0.4, wet
curing should be used, but only if it can be applied thoroughly and continuously. If
such an assurance is not possible, then membrane curing is preferable, but that, too,
has to be well executed.
It is obvious that the membrane must be continuous and undamaged. The timing of
spraying is also critical. The curing spray should be applied after bleeding has stopped
bringing water to the surface of the concrete but before the surface has dried out: the
optimum time is the instant when the free water on the surface of the concrete has
disappeared so that the water sheen is no longer visible. However, if bleeding has not
stopped, the curing membrane should not be applied even if the surface of the
concrete appears dry in consequence of a high rate of evaporation. For this purpose,
an evaporation rate of 1 kg/m2 per hour can be taken as ‘high’. The rate can be
calculated using Figs 7.2 to 7.5, based on Lerch’s results;7.37 alternatively, a chart in
ACI 308R-86, based on the same source7.37 as these figures, can be used.
When a high rate of evaporation removes water faster than it is brought up by
bleeding, Mather7.6 recommends wetting the concrete and delaying the application of
the curing compound until bleeding has ceased.
Some concretes, for instance those containing silica fume, exhibit no bleeding, in
which case the curing membrane should be applied without delay. If the curing
compound is applied to a surface which has dried out, the spray will penetrate into the
concrete and prevent further hydration within the outer zone. Moreover, no effective
continuous membrane would be formed.7.6

352
 With slip-forming, where the form is effectively removed after several hours, the
immediate application of curing is important if there are durability requirements, or
for strength reasons in the case of thin members. On the other hand, ordinary
formwork left in place is a means of preventing the loss of moisture from vertical
surfaces. After it has been loosened, application of water is possible.
Tests on curing compounds
The efficiency of curing compounds in terms of the extent to which they permit the
loss of water from the surface of a standard mortar can be determined by tests. British
Standard BS 7542 : 1992 uses 1:3 mortar with a water/cement ratio of 0.44, and
exposure to 38 °C (100 °F) and a relative humidity of 35 per cent for 72 hours; the
percentage reduction in the loss of water compared with a specimen without a
membrane is taken to represent the curing efficiency. The test method of ASTM C
156-09a is similar but the performance of the compound is expressed as a loss of
water per unit area. The reproducibility of this test is said to be poor.7.4
Neither the British nor the American test measures the quality of the cured
concrete in the surface zone, which is what is of interest in practice but is not easy to
determine. Various proposed other tests are too cumbersome for practical use or else
interfere with the concrete being tested.
In tests, the surface of the mortar is level and finished with a float. In practice, the
surface of the concrete may be coarsely brushed or tined (as in the case of highway
slabs) and this affects the amount of curing compound necessary. Also, because a
uniform and continuous membrane is more difficult to achieve under such
circumstances, a good water-retaining performance in the test may fail to be matched
in practice.
Length of curing
The period of curing required in practice cannot be prescribed in a simple manner: the
relevant factors include the severity of the drying conditions and the expected
durability requirements. As an example, the minimum periods of curing for external
exposure, including freezing and thawing but not the use of de-icing agents and for
exposure to aggressive chemicals are given in Table 7.1, derived from BS EN 206-1 :
2007. If concrete is to be subjected to abrasion, doubling of the curing periods is
desirable. Minimum periods of curing are given in BS 8110-1 : 1997.

Table 7.1. Minimum Curing Times (in days) Recommended in BS EN 206-1 :

2007

353
 The requirements for striking formwork are governed by the strength of concrete.
This can be estimated from the maturity of the concrete (see p. 306) or by tests on
companion compressive strength specimens (see p. 584) or else by non-destructive
tests. Guidance is given by Harrison.7.8
It was stated earlier that curing should start at the earliest possible instant and
should be continuous. Occasionally, intermittent curing is applied, and it is useful to
appreciate its effect. In the case of concrete with a low water/cement ratio, continuous
curing at an early age is vital as partial hydration may make the capillaries
discontinuous: on renewal of curing, water would not be able to enter the interior of
the concrete and no further hydration would result. However, mixes with a high
water/cement ratio always retain a large volume of capillaries so that curing can be
effectively resumed at any time, but the earlier the better.
The preceding discussion has laid much stress on the importance of proper curing.
Curing is always specified but rarely adequately executed. And yet, inadequate curing
is responsible for a great many durability problems with concrete, especially
reinforced concrete. For this reason, the importance of curing cannot be
overemphasized.
Autogenous healing
Fine cracks in fractured concrete, if allowed to close without tangential displacement,
will heal completely under moist conditions. This is known as autogenous healing,
and is due primarily to the hydration of the hitherto unhydrated cement which
becomes exposed to water upon the opening of the cracks. Healing is also aided by
the formation of insoluble calcium carbonate from the calcium hydroxide in hydrated
cement if carbonation takes place. Some mechanical blocking of the cracks may also
occur if very fine material is suspended in the water.
The maximum width of cracks which can undergo autogenous healing is estimated
to be between 0.1 and 0.2 mm, and the necessary moist conditions include frequent
periodic wetting as well as immersion,7.28 but not fast-flowing water or high water
pressure, which is not conducive to reducing the movement of water through the crack.
The application of pressure across the crack assists in healing.
In young concrete, cracks 0.1 mm wide can heal after several days but 0.2 mm
wide cracks require several weeks.7.28 Generally, the younger the concrete, i.e. the
more unhydrated cement it contains, the higher the re-gain of strength, but healing

354
without a loss of strength has been observed at ages up to three years. It has been
reported7.31 that, even when healed, cracks represent a weakness zone in which
renewed cracking may occur under future adverse conditions.
Means of encouraging autogenous healing by incorporation of suitable bacteria in
the mix are being studied in laboratory tests (see p. 266).
Variability of strength of cement
Up to now we have not considered the strength of cement as a variable in the strength
of concrete. By this we do not mean the differences in the strength-producing
properties of cements of different types, but the variation between cements of
nominally the same type: they vary fairly widely, and it is this variation that is
considered in this section.
The strength requirements for cement were discussed in Chapter 2. Traditionally,
only a minimum strength at certain ages has been prescribed, so that there should be
no objection to a cement with a much higher strength. The cement manufacturers
advance this argument forcefully and are unsympathetic to cement users who want to
take economic advantage of the actual higher strength of cement and who complain
when, at some instant, the strength margin above the specified minimum is
substantially reduced.
One consequence of the absence of an upper limit on strength is that there is an
overlap in the strengths of Type I and Type III cements: occasionally Type I cements
have been found to have strengths as high as twice the specified minimum.7.41
The absence of a maximum specified strength persists in most specifications.
However, European Standard BS EN 197-1 : 2000, BS 12 : 1996. German standards
(which pioneered this approach) prescribe a maximum strength for most of the cement
classes at a value 20 MPa greater than the minimum. This range of strengths for a
given class of cement is high, although probably justified economically for a mass
product with a wide range of uses.
The variation in strength of cement is due largely to the lack of uniformity in the
raw materials used in its manufacture, not only between different sources of supply,
but also within a pit or quarry. Furthermore, differences in details of the processes of
manufacture and, above all, the variation in the ash content of coal used to fire the
kiln, contribute to the variation in the properties of commercial cements. This is not to
deny that the modern manufacture of cement is a highly sophisticated process.
Pioneer work in the variation in the strength of cement by Walker and Bloem7.42 has
contributed to the development of a test method for the evaluation of cement strength
uniformity from a single source, ASTM C 917-05. This method uses the mortar cube
strength test ASTM C 109-08 and relies on a moving average of five grab (spot)
samples. An example of the variability in a single plant over a period of three years is
given in Fig. 7.9. It can be seen that there was a reduction in variability between 1982
and 1984; the standard deviation* of the 7-day strength at the end of the period was
1.4 MPa (208 psi). Tests7.14 at 87 United States cement plants, conducted in 1991,
showed that 81 per cent of them had a standard deviation of 7-day strength lower than
2.10 MPa (300 psi); at 28 days, only 43 per cent of the plants had a standard deviation
lower than 2.10 MPa. The increase in the standard deviation with age is typical of
American cements7.12 but not necessarily of cements made elsewhere.
*
The statistical terms are defined on p. 642.

355
Fig. 7.9. Plot of moving averages of strength of five tests on 28-day mortar cubes
(made to ASTM C 109) using cement from a single plant in the years 1982 to
1984 (based on ref. 7.13)
The large range of strengths of cement from a single plant in Fig. 7.9 should be
noted: a range of 28-day strength of 7 MPa (1000 psi) in a period of a few months is
not uncommon. Clearly, using cement with a smaller and known variability would
result in an economic advantage over relying on the minimum strength. There remains,
however, the problem of the relatively poor precision of the ASTM C 109-08 mortar
test used to measure the strength of cement. Nevertheless, large purchasers of cement
can influence its variability by requiring testing to ASTM C 917-05 and agreeing
appropriate limits.
It is important to be clear about the use of grab samples and of a moving average.
Values of single grab samples might not be representative and would be unduly
affected by testing errors. On the other hand, composite samples, which are obtained
by putting together sub-samples from production during 24 hours, give an unduly
smoothed-out result.
What is the relevance of the strength of cement to the strength of concrete made
with that cement? It is rational to expect a direct influence7.78 (see Fig. 7.10), even
though many other factors also affect the strength of concrete. This relation between
cement and concrete strengths may seem obvious, but in the past it was claimed7.32 that
there is no correlation between the strength of concrete and the strength of cement, as
established by the cement manufacturers’ testing, used in making that concrete.

356
Fig. 7.10. Moving average strength of mortar cubes (made to ASTM C 109) and
average strength of concrete cylinders at the age of 28 days in the period March
to July 1980 (based on ref. 7.78). Note: The ordinates for mortar and concrete
are not the same; the two plots have been moved close to one another
This kind of argument misses the crucial point that a composite sample of cement
obtained over a 24-hour period represents average properties of the thousands of
tonnes of cement produced during that period. Inevitably, there are variations within
that bulk of cement, only a very small part of which is used in making a given batch
of concrete. At the same time, making concrete also introduces variability.
As an aside, it may be useful to make a comment about the use of cement
manufacturers’ test certificates in research. Often, the properties of cement such as
chemical composition reported in the test certificate are used by the researcher as a
test parameter. If the test certificate refers to the average of a 24-hour production, the
properties as listed cannot be considered as necessarily applying to the actual cement
used by the researcher. If they are so considered, spurious correlations with the
property investigated may be found; alternatively, the experimental work may fail to
show a real correlation through no fault of the researcher.7.33
It must be realized that the use of admixtures would clearly upset the relationship
between cement and concrete strengths because the precise influence of admixtures
depends on the properties of the cement used, whereas the cement strength test uses a
mortar without any admixtures.
With the introduction of performance-based specifications for cement, it is
important to know more about its true strength characteristics, which must affect the
strength of the concrete made with a given cement. The situation becomes more
complicated when cement comes from different sources.
The variation in the strength of cement from different plants is obviously much
larger than when a single plant is the supplier. Table 7.2 gives the data for 87 plants in
the United States, tested in 1991;7.14 the strengths are those of mortar cubes according
to ASTM C 109-08. It must not be forgotten, however, that variation in cement
accounts, at the most, for one-half of the variation in the strength of site test
specimens; U.S. Bureau of Reclamation data7.57 suggest a typical value of one-third.
The variation in the strength of site cubes is discussed on p. 639. More recent studies
on the variability of strength of cement can be found in ref 7.102.

357
Table 7.2. Strength of Cement Produced in 87 American Plants in 19917.14 (shown
as a percentage of plants with average strength lower than indicated) (Copyright
ASTM–reproduced with permission)

Finally, it should be stressed that the variation in cement affects to the greatest
extent the early strength of concrete, i.e. the strength most often determined by test
but not necessarily the strength of greatest practical significance. Furthermore,
strength is not the only important characteristic of concrete: from considerations of
durability and permeability, a cement content in excess of that needed for strength
may well be required, in which case the variability of cement becomes unimportant.
Changes in the properties of cement
In the preceding section, we considered the variation in the strength of cement
produced in a single plant over a period of several months or a year. Some reference
was also made to the differences in the strength of cements made in different plants
during a single year. There is, in addition, a systematic change in the strength of
cement with time. Indeed, there has been such a continuing change over many years
in consequence of the improvement in the manufacture of cement7.10,7.39 (see Fig. 7.11).

358
Fig. 7.11. Changes in gain of strength of cements with age between 1916 and the
1990s (measured on standard cylinders of concrete with a water/cement ratio of
0.53 (based on refs 7.10 and 7.39 and private data)
First of all, we can give an example7.1 of the difference in the average properties of
cements produced in 1923 and in 1937. Two series of tests spanning a 50-year life of
concrete stored outdoors in Wisconsin, United States, gave data on the strength
development. The 1923 concretes were made with cements with a high C2S content
and a low fineness: their compressive strength increased in proportion to the
logarithm of age up to 25 or 50 years. The concretes made in 1937 used cements with
a lower C2S content and a high fineness: their compressive strength increased in
proportion to the logarithm of age for about 10 years but, thereafter, decreased or
remained constant.7.1 This change in behaviour is mainly of historical interest but it
helps to understand the differences in the behaviour of concretes of various ages.
A more recent change, namely that around the 1960s, merits particular attention
because it has had far-reaching consequences for concrete production practice.
The changes in British cements are well documented7.16,7.21 but they occurred in other
countries as well. The change of greatest practical interest was the increase in the 28-
day strength, and also in the 7-day strength, of mortar made with a fixed water/cement
ratio. The main reason for this was a large increase in the average content of C3S:
from about 47 per cent in 1960 to about 54 per cent in the 1970s.7.16 There was a
corresponding decrease in the content of C2S so that the total content of calcium
silicates remained constant at 70 to 71 per cent. This change was made possible by
improvements in the methods of manufacture of cement, but it was also driven by the
benefits of using a ‘stronger’ cement as perceived by the users, namely: reduction in
cement content for a given specified strength, earlier removal of formwork, and faster
construction. Such benefits were, unfortunately, associated with disadvantages.

359
 There was no significant change in the fineness of cement, which is not surprising
because of the high cost of grinding clinker.7.16,7.20
The high rate of increase in strength up to 7 days and the rate of increase between
7 and 28 days have changed in consequence of the higher alkali content in modern
cements as well as because of the change in the ratio of C3S to C2S. The ratio of the
strength at 28 days to that at 7 days has decreased substantially. For concrete with a
water/cement ratio of 0.6, a decrease in the 28- to 7-day strength ratio from about 1.6
prior to 1950 to about 1.3 in the 1980s was reported;7.20 these figures are only
examples of the behaviour of some British cements and are not necessarily generally
valid. At lower water/cement ratios, the ratio of the 28-day strength to the 7-day
strength is lower. Likewise, the increase in strength beyond the age of 28 days is
much reduced when modern cements are used so that it should no longer be relied
upon in the design of structures which will be subjected to full load only at an
advanced age.
An example of the change in the 28-day strength of cement between 1970 and
1984 is shown in Fig. 7.12.7.21 It can be seen that concrete with a characteristic cube
strength (see p. 732) of 32.5 MPa (4700 psi), which in 1970 required a water/cement
ratio of 0.50, could be achieved in 1984, using a water/cement ratio of 0.57.
Assuming that, for the workability to remain constant, the same water content of, say,
175 kg per cubic metre of concrete was maintained, it was possible to reduce the
cement content from 350 to 307 kg/m3.

360
 Fig. 7.12. Relation between the characteristic strength of concrete and
water/cement ratio for concretes made in 1970 and in 1984; 20 mm ( in.)
maximum aggregate size, slump of 50 mm (2 in.) (based on ref. 7.21)
More generally, over the longer period between the 1950s and the 1980s, for
concrete of a given strength and workability, it was possible to reduce the cement
content by 60 to 100 kg per cubic metre of concrete and concomitantly to increase the
water/cement ratio by between 0.09 and 0.13.7.20
While a higher 28-day strength of concrete at a given water/cement ratio could be
economically exploited, there were consequential disadvantages. Concrete having the
same 28-day strength as before (when the ‘old’ cements were used) can be made
using a higher water/cement ratio and a lower cement content, as shown in the
preceding paragraph. Both these concomitant changes result in concrete with a higher
permeability and therefore more liable to carbonation and penetration by aggressive
agents, and generally of lower durability.

361
 Moreover, the absence of a significant increase in strength beyond the age of 28
days7.20,7.21 removed a long-term improvement in concrete which had reassured users in
the past (even if such improvement was not taken into account in design).
The rapid early gain in strength also means that strengths adequate for removal of
formwork are achieved earlier than was the case with the ‘old’ cements so that
effective curing ceases at an early age.7.17 The adverse consequences of this were
discussed earlier in this chapter.
These consequences were not foreseen, partly because many concrete users were
preoccupied with exploiting the high early strength properties of cement, and partly
because the concrete specifications were couched predominantly in terms of a 28-day
strength, which remained the same as it had been when the ‘old’ cements were used.
Although the above data refer to British cements, the changes have occurred
worldwide, albeit not at the same time, the driving force being the modernization of
cement plants. French figures may be of interest: between the mid-1960s and 1989,
the average C3S content of Portland cement increased from 42 to 58.4 per cent, with
C2S decreasing simultaneously from 28 to 13 per cent.7.15
The increase in the average 28-day strength appears to continue. In the United
States, between 1977 and 1991, the strength of mortar made according to ASTM C
109-93 increased from 37.8 MPa (5470 psi) to 41.5 MPa (6020 psi).7.14
Fatigue strength of concrete
In Chapter 6, we considered only the strength of concrete under static loading. In
many structures, however, repeated loading is applied. Typical of these are offshore
structures subjected to wave and wind loading, bridges, road and airfield pavements,
and railway sleepers (ties); the number of cycles of loading applied during the life of
the structure may be as high as 10 million, and occasionally even 50 million.
When a material fails under a number of repeated loads, each smaller than the
static compressive strength, failure in fatigue is said to take place. Both concrete and
steel possess the characteristics of fatigue failure but, in this book, the behaviour of
concrete alone is dealt with.
Let us consider a concrete specimen subjected to alternations of compressive stress
between values σ1 (≥0) and σh (>σ1). The stress–strain curve varies with the number of
load repetitions, changing from concave towards the strain axis (with a hysteresis loop
on unloading) to a straight line, which shifts at a decreasing rate (i.e. there is some
irrecoverable deformation) and eventually becomes concave towards the stress axis.
The degree of this latter concavity is an indication of how near the concrete is to
failure. Failure will, however, take place only above a certain limiting value of σh,
known as fatigue limit or endurance limit. If σh is below the fatigue limit, the stress–
strain curve will indefinitely remain straight, and failure in fatigue will not take place.
The changes in the stress–strain curve with the number of applied cycles are
illustrated in Fig. 7.13 for compressive loading and in Fig. 7.14 for direct tension.7.94

362
 Fig. 7.13. Stress–strain relation of concrete under cyclic compressive loading

363
Fig. 7.14. Stress–strain relation of concrete under cyclic loading in direct tension
(based on ref. 7.94)
The change in strain with the number of cycles of loading can be described as
consisting of three phases.7.83 In Phase 1, that is, the initiation phase, strain increases
rapidly, but at a progressively decreasing rate, with the number of cycles of loading.
In Phase 2, which represents the stable state, strain increases approximately linearly
with the number of cycles. In Phase 3, which represents instability, strain increases at
a progressively increasing rate until failure in fatigue takes place. An example of this
behaviour is shown in Fig. 7.15.

364
 Fig. 7.15. Relation between strain and relative number of cycles of loading in
compression, expressed as proportion of number of cycles to failure (maximum
stress equal to 0.75 of the static strength; minimum stress equal to 0.05 of the
static strength) (based on ref. 7.83)
If the stress–strain curve for unloading were also drawn in Fig. 7.13, a hysteresis
loop in each cycle could be seen. The area of this loop decreases with each successive
cycle and then eventually increases prior to fatigue failure.7.43 There does not seem to
be such an increase in specimens which do not fail in fatigue. If we plot the area of
each successive hysteresis loop as a percentage of the area of the first loop, the
variation with the number of cycles is as shown in Fig. 7.16.

365
 Fig. 7.16. Variation in the area of the hysteresis loop as a percentage of the first
loop with the number of cycles7.43
The interest in the hysteresis loop arises from the fact that its area represents the
irreversible energy of deformation, and is manifested by a rise in temperature of the
specimen. The irreversible deformation involved is probably in the form of
microcracking. Pulse velocity measurements have shown7.43 that it is the development
of cracks that is responsible for the change in behaviour near failure.
The strain at failure in fatigue is much larger than in static failure and can be as
high as 4 × 10–3 after 13 million cycles at 3 Hz. Generally, the specimen with a longer
fatigue life has a higher non-elastic strain at failure (Fig. 7.17).

Fig. 7.17. Relation between non-elastic strain near failure and number of cycles
at failure7.43
The elastic strain also increases progressively with cycling. This is shown in Fig.
7.18 by the reduction in the secant modulus of elasticity (see p. 414) with an increase
in the percentage of the ‘fatigue life’ used up. This relation is independent of the level
of stress in the fatigue test and is, therefore, of interest in assessing the remaining
fatigue life of a given concrete.

366
 Fig. 7.18. Relation between the ratio of the secant modulus of elasticity at the
given instant (E) to the modulus at the beginning of cycling (E0) and the
percentage of fatigue life used up7.43
The lateral strain is also affected by the progress of cyclic loading, the Poisson’s
ratio decreasing progressively.
Cyclic loading below the fatigue limit improves the fatigue strength of concrete, i.e.
concrete loaded a number of times below its fatigue limit will, when subsequently
loaded above the limit, exhibit a higher fatigue strength than concrete which had
never been subjected to the initial cycles. The former concrete also exhibits a higher
static strength by some 5 to 15 per cent, but a value as high as 39 per cent has been
reported.7.85 It is probable that this increase in strength is due to a densification of
concrete caused by the initial low-stress level cycling, in a manner similar to
improvement in strength under moderate sustained loading.7.45 This property is akin to
strain hardening in metals, and is of particular interest because concrete under static
loading is a strain-softening rather than strain-hardening material.

367
 Strictly speaking, concrete does not appear to have a fatigue limit, i.e. a fatigue
strength at an infinite number of cycles (except when stress reversal takes place). It is
usual, therefore, to refer to fatigue strength at a very large number of cycles, such as
10 million, but for some sea structures an even higher number may be appropriate.
The fatigue strength can be represented by means of a modified Goodman diagram
(see Fig. 7.19). The ordinate from a line at 45° through the origin shows the range of
stress (σh – σ1) for a given number of cycles; σ1 is generally greater than zero (arising
from the dead load) while σh is due to the dead plus live (transient) load. Thus, the
range of stress that a given concrete can withstand a specified number of cycles can
be read off the diagram. For a given σ1, the number of cycles is very sensitive to the
range of stress. For instance, an increase in range from 57.5 to 65 per cent of the
ultimate static strength has been found to decrease the number of cycles by a factor of
40.7.46

Fig. 7.19. Modified Goodman diagram for concrete in compression fatigue (N is

number of cycles)
The modified Goodman diagram (see Fig. 7.19) shows that, for a constant range of
stress, the higher the value of the minimum stress the lower the number of cycles that
a given concrete can withstand. This is of significance in relation to the dead load of a
concrete member which is to carry a transient load of a certain magnitude.
From the fact that the lines of Fig. 7.19 rise to the right, it can also be seen that the
fatigue strength of concrete is lower the higher the ratio σh/σ1.
The frequency of the alternating load, at least within the limits of 1.2 to 33 Hz,
does not affect the resulting fatigue strength;7.47 higher frequency is of little practical
significance. This applies both in compression and in flexure, the similarity between
fatigue behaviour in the two types of loading, as well as in splitting
tension,7.63 suggesting that the failure mechanism is the same.7.48 In fact, the fatigue
behaviour in flexure parallels closely that in compression (Fig. 7.20). The fatigue

368
strength in flexure (for 10 million cycles) was found to be 55 per cent of the static
strength;7.84 values of 64 to 72 per cent have also been reported.7.99 By comparison, in
compression, the fatigue strength was reported to be between 60 and 64 per cent after
the same number of cycles, but a value of 55 per cent has also been quoted.7.85 Because
of a high scatter in the fatigue test results, the application of the concept of probability
of survival in fatigue has to be used in design.7.95

Fig. 7.20. Modified Goodman diagram for concrete in flexure fatigue7.44

Some tests have shown that lateral pressure increases the fatigue life of concrete,
but not at very high stresses.7.58 Generally, the pattern of fatigue behaviour of plate-
shaped specimens in biaxial compression is broadly similar to that under uniaxial
compression; compressive lateral stress of 0.2 and 0.5 of the axial stress was found to
increase the fatigue life by up to 50 per cent compared with that under uniaxial
compression.7.87 An increase in the fatigue life of cubes under biaxial compression was
also reported.7.96 The reason for this is probably the fact that a compressive lateral
stress restrains the development of microcracking which is responsible for fatigue
failure. This observation is of interest as in many structural situations lateral
compression is present.
Some tests have shown that the moisture condition of concrete prior to loading
affects its fatigue strength in flexure: oven-dried specimens show the highest strength
and partially dried ones the lowest; wet specimens are in between (Fig. 7.21). The
explanation of this behaviour lies in differential strains induced by the moisture
gradient.7.59 The apparent effect is thus test-related. Submersion in water does not
affect the fatigue life.7.86

369
 Fig. 7.21. Effect of moisture condition on fatigue performance of concrete
specimens7.59 (Crown copyright)
Generally speaking, the ratio of fatigue strength to static strength is independent of
the water/cement ratio, the cement content, type of aggregate, and age at loading
because these factors affect both the static and the fatigue strength in the same manner.
As strength increases with age, fatigue strength both in compression and in flexure
also increases.7.63 The important point is that, at a given number of cycles, fatigue
failure occurs at the same fraction of ultimate strength, and is thus independent of the
magnitude of this strength (both in compression and in splitting tension)7.64 and of the
age of concrete,7.47 although some tests suggest an increase in fatigue life with age.7.59 It
can thus be seen that a single parameter is critical in fatigue failure.
Murdock7.47 expressed the view that the deterioration of bond between the hydrated
cement paste and aggregate is responsible for this failure. Tests have shown that
fatigue specimens had fewer broken aggregate particles than specimens which failed
in a static test.7.49 Thus, failure at the aggregate–paste interface is probably dominant in
fatigue; in mortar, fatigue failure is believed to take place at the interface of the fine
aggregate particles.7.43 It is likely that a smaller maximum size of aggregate leads to a
higher fatigue strength,7.60 probably because of greater homogeneity of concrete.
Air-entrained concrete and lightweight aggregate concrete have the same fatigue
behaviour as concrete made with ordinary aggregate,7.50,7.61,7.86 although air entrainment
may reduce the fatigue life in flexure.7.98 Fatigue in concrete cylinders occurs in the
same way as in large specimens subjected to low-frequency loading.7.62

370
 High strength concrete also exhibits behaviour similar to ordinary concrete, but
shows a lower deformation (probably due to a higher modulus of elasticity) and a
higher fatigue life under high values of maximum stress.7.83 The performance of high
strength concrete in fatigue can thus be considered to be good, but failure is rather
sudden.7.83
The fatigue strength of concrete is increased by rest periods (this does not apply
when there are stress reversals), the increase being proportional to their duration
between 1 and 5 minutes; beyond the 5-minute limit there is no further increase in
strength. With the rest periods at their maximum effective duration, their frequency
determines the beneficial effect.7.47 The increase in strength caused by rest periods is
probably due to relaxation of concrete (primary bonds, which remained intact,
restoring the internal structure to its original configuration), as evidenced by a
decrease in the total strain; this decrease occurs rapidly after the cessation of cycling.
Murdock7.47 suggested that fatigue failure occurs at a constant strain, independent of
the applied stress level or of the number of cycles necessary to produce failure. This
behaviour of concrete would add further support to the concept of ultimate strain as
failure criterion.
Most fatigue tests are conducted under cyclic loading of constant shape. However,
structures such as those subjected to wave action undergo variable amplitude loading.
Tests involving variable stress levels have shown that the sequence of low-stress and
high-stress cycling affects the fatigue life. In particular, if high-stress cycling
succeeds the low-stress cycling, the fatigue strength is reduced. It follows that Miner’s
hypothesis7.88 of linear accumulation of damage (valid for metals) does not apply to
concrete7.44,7.65,7.89 and may err on the unsafe side. A modification of Miner’s hypothesis
which takes into account the sequence of variable amplitude loading was developed
by Oh;7.100 its general validity is still to be established.
It should also be noted that, for a given maximum stress in the cycle, as the
amplitude of stresses decreases, we are no longer dealing with fatigue but rather with
sustained loading which leads to creep failure (see p. 474). The duration of cycling
becomes therefore important. Expressions taking this into account were developed by
Hsu,7.90 who considers that separate equations for fatigue life are needed for low-cycle
loading of the type caused by earthquakes; direct application of test results from
laboratory tests at high frequency may be unsafe.7.97
While this book is not concerned with the fatigue behaviour of reinforced and
prestressed concrete, we should note that fatigue cracks in concrete act as stress-
raisers, thus magnifying the vulnerability of the steel to fatigue failure7.51 (if the stress
in it is in excess of its critical fatigue stress value).
Another observation relevant to reinforced concrete is that the fatigue strength of
bond of concrete with the reinforcement is the governing factor in reinforced concrete
subjected to cyclic loading.7.86 As bond is improved by the incorporation of silica fume
in the mix, this would explain why the presence of silica fume in high strength
lightweight aggregate concrete increases the fatigue strength of reinforced concrete
members, compared with members made with concrete of the same strength but
without silica fume.
It is possible that the fatigue of bond with reinforcement is best expressed in terms
of cumulative deformation (slip) in a static bond test.7.82
Impact strength

371
Impact strength is of importance when concrete is subjected to a repeated falling
object, as in pile driving, or a single impact of a large mass at a high velocity. The
principal criteria are the ability of a specimen to withstand repeated blows and to
absorb energy.
Green7.52 studied the number of blows of a ballistic pendulum which 100 mm (4 in.)
concrete cubes can withstand before reaching the no-rebound condition, this stage
indicating a definite state of damage. He found that impact tests on compression
specimens, when conducted with a small hammer (25 mm (1 in.) diameter face), lead
to a greater scatter of results than tests on static compressive strength of the concrete.
This arises from the fact that, in the standard compression test, some relief of a highly
stressed weak zone is possible due to creep, whereas in the impact test no
redistribution of stresses is possible during the very short period of deformation.
Hence, local weaknesses have a greater influence on the recorded strength of a
specimen.
In general, the impact strength of concrete increases with an increase in
compressive strength,7.92 but the higher the static compressive strength of the concrete
the lower the energy absorbed per blow before cracking.7.52
Figure 7.22 gives some examples of the relation between the impact strength and
the compressive strength.7.52 It can be seen that the relation is different for each coarse
aggregate and storage condition of the concrete. For the same compressive strength,
the impact strength is greater for coarse aggregate of greater angularity and surface
roughness. This observation was confirmed by Dahms7.66 and supports the
suggestion7.53 that impact strength is more closely related to the tensile strength of
concrete than to its compressive strength. Thus, concrete made with a gravel coarse
aggregate has a low impact strength, failure taking place due to insufficient bond
between mortar and coarse aggregate. On the other hand, when the surface of the
aggregate is rough, the concrete is able to develop the full strength of much of the
aggregate in the region of failure.

372
 Fig. 7.22. Relation between compressive strength and number of blows to ‘no-
rebound’ for concretes made with different aggregates and Type I cement, stored
in water7.52
A smaller maximum size of aggregate significantly improves impact strength both
in compression7.66 and in splitting tension.7.93 Impact strength in compression is
improved by the use of aggregate with a low modulus of elasticity and a low
Poisson’s ratio.7.66 Cement content below 400 kg/m3 (670 lb/yd3) is
advantageous.7.66 The influence of fine aggregate is not well defined but the use of fine
sand usually leads to a slightly lower impact strength. Dahms7.66 found a high content
of sand advantageous. We could try to generalize and say that a mix of materials
which have a limited variation in properties is conducive to a good impact strength.
Extensive tests on the impact strength of concretes with different properties were
made by Hughes and Gregory.7.54
Storage conditions influence the impact strength in a manner different from
compressive strength. Specifically, the impact strength of water-stored concrete is
lower than when the concrete is dry, although the former concrete can withstand more
blows before cracking. Thus, as already stated, the compressive strength without
reference to storage conditions, does not give a satisfactory indication of the impact
strength.7.52
Repeated impact tests on slabs have also been used,7.92 the end point being
perforation of the slab. Such tests are usually directed towards a direct structural

373
application and often involve fibre-reinforced concrete. Impact tests in splitting
tension can also be performed.
There is evidence that, under uniformly applied impact loading (a condition
difficult to achieve in practice), the impact strength of concrete is significantly greater
than its static compressive strength. This increase in strength would explain the
greater ability of concrete to absorb strain energy under uniform impact. Figure
7.23 shows that strength increases greatly when the rate of application of stress
exceeds about 500 GPa/s, reaching, at 4.9 TPa/s, more than double the value at
normal speeds of loading (about 0.5 MPa/s).7.67 Impact at a loading rate six orders of
magnitude greater than in a static test led to a 50 per cent increase above the static
compressive strength.7.91 In splitting tension, the same increase in the loading rate was
found to result in an 80 per cent increase above the static strength.7.93

Fig. 7.23. Relation between compressive strength and rate of loading up to

impact level7.67
The influence of the rate of application of strain on compressive strength is shown
in Fig. 7.24. It can be seen that, at very high strain rates, there is a large increase in
compressive strength, probably due to inertial resistance of concrete to
microcracking;7.80 at low rates, the effect of creep may be dominant. The influence of
the strain rate upon the tensile strength of concrete is even larger,7.81 the free water in
the hardened cement paste playing a significant role.7.79 The subject of the influence of

374
the rate of loading on strength is considered also in Chapter 12 in connection with
testing.

Fig. 7.24. Relation between relative increase in compressive strength (as a

proportion of static strength) and the strain rate for concretes of different
strengths (based on ref. 7.80)
Electrical properties of concrete
Electrical properties are of concern in some specific applications such as railway ties
(sleepers) (where inadequate resistivity affects some signalling systems) or in
structures in which concrete is used for protection from stray currents. Electrical
resistance of concrete also influences progress of corrosion of embedded steel.
Electrical properties are also of interest in studies of the properties of both fresh and
hardened concrete.
In the vicinity of underground cables, concrete may be subjected to impressed
electrical activity but, under the usual operating conditions, concrete offers a high
resistance to the passage of electric current to or from embedded steel. This is largely
due to the electro-chemical effect which concrete has on steel in contact with it,
arising from the alkalinity of the electrolyte within the concrete. Such a protection
applies within the potential range of about +0.6 to –1.0 V (with respect to a copper

375
sulfate electrode), the current being primarily controlled by polarization effects and
not by the ohmic resistance of concrete.7.69
Moist concrete behaves essentially as an electrolyte with resistivity of up to about
100 ohm-m; this is within the range of semi-conductors. Air-dried concrete has a
resistivity of the order of 104 ohm-m.7.19 On the other hand, oven-dry concrete has a
resistivity of about 109 ohm-m, which means that such concrete is a good
insulator.7.70 The insulating or dielectric properties have been studied by Halabe et al.7.27
This large increase in resistivity of concrete on removal of water is interpreted to
mean that electric current is conducted through moist concrete essentially by
electrolytic means, that is by ions in the evaporable water. However, when the
capillaries are segmented, passage of the electric current through gel water takes place.
The resistivity of normal aggregate is infinitely larger. For concrete of given mix
proportions, drying out in the air increases the resistivity of the surface zone. For
instance, at a water/cement ratio of 0.50, Tritthart and Geymayer7.34 reported an
eleven-fold increase; the increase was even larger at higher water/cement ratios.
It can therefore be expected that any increase in the volume of water and in
concentration of ions present in the pore water decreases the resistivity of cement
paste, and indeed resistivity decreases sharply with an increase in the water/cement
ratio. This is shown in Table 7.3 for hydrated cement paste and in Fig. 7.25 for
concrete. A decrease in the cement content of the concrete also results in an increased
resistivity7.18 because, at a constant water/cement ratio but a lower cement content,
there is less electrolyte available for the current to pass.

Table 7.3. Influence of Water/Cement Ratio and Length of Moist Curing on

Resistivity of Cement Paste7.70

376
 Fig. 7.25. Relation between electrical resistivity and water/cement ratio for
concrete with a maximum size of aggregate of 40 mm ( in.) made with
ordinary (Type I) Portland cement, tested at the age of 28 days (based on
ref. 7.18)
The resistivity of concretes of varying composition is given by Hughes et al.7.18 If
necessary, the values of resistivity of hydrated cement paste can be converted into
resistivity of concrete which includes this paste, approximately in an inverse ratio of
the relative volume of hydrated cement paste.7.19

377
 The long-term reactions involving ground granulated blastfurnace slag in concrete
cause a continuing increase in electrical resistivity. This can be as much as an order of
magnitude, compared with concrete containing Portland cement only.7.30 Silica fume
also increases the resistivity. The effects of ground granulated blastfurnace slag and
silica fume are of significance when the progress of corrosion of steel reinforcement
is controlled by the electrical resistance of concrete (see Chapter 11).
Like other ions in the pore water, chlorides greatly reduce the resistivity of
concrete and mortar; for the latter, a 15-fold decrease was reported.7.71 The influence
on resistivity of salinity of the mixing water is greatest in concrete with high
water/cement ratios and is quite small in high strength concrete.7.72
During the first few hours after mixing, the resistivity of concrete increases very
slowly, then increases rapidly up to the age of about 1 day, and thereafter increases at
a reduced rate or becomes constant7.18 unless the concrete dries out; drying increases
the resistivity.
The resistivity of concrete immersed in sea water can become greatly increased by
the formation of a thin surface layer of magnesium hydroxide and calcium
carbonate.7.101 If this layer is removed, the resistivity is the same as for storage in fresh
water.
The relation between resistivity of concrete and the volume fraction occupied by
water can be derived from the laws of conductivity of heterogeneous conductors.
However, for the range of the usual concrete mixes, the water content varies
comparatively little for a given aggregate grading and workability, and the resistivity
becomes more dependent on the cement used7.73 because the chemical composition of
the cement controls the quantity of ions present in the evaporable water. Some idea of
the influence of cement on resistivity can be obtained from Table 7.4, from which it
can be seen that the resistivity of concrete made with high-alumina cement is 10 to 15
times higher than when Portland cement in the same proportions is used7.73 (see Fig.
7.26).

Table 7.4. Typical Electrical Properties of Concrete (based on ref. 7.74)

378
379
 Fig. 7.26. Relation between resistivity and applied voltage for a 1:2:4 concrete
with a water/cement ratio of 0.49, oven-dried and cooled in a desiccator7.74
Admixtures generally do not reduce the resistivity of concrete.7.70 However, special
additions can be used to vary the resistivity. For instance, the addition to concrete of
finely divided bituminous material, with subsequent heat treatment at 138 °C (280 °F),
increases the resistivity, especially under wet conditions.7.75 Conversely, in cases where
static electricity is undesirable and a decrease in the insulation resistance of concrete
is required, satisfactory results can be achieved by the addition of acetylene carbon
black (2 to 3 per cent by mass of cement).7.75 Electrically conductive concrete can be
obtained by replacing fine aggregate with a granulated conductive aggregate
consisting of almost pure crystalline carbon prepared as a proprietary product.
Resistivity is between 0.005 and 0.2 ohm-m; the compressive strength and other
properties are reported not to be significantly affected.7.76
The resistivity of concrete increases with an increase in voltage.7.74 Figure
7.26 illustrates this relation for oven-dried specimens not allowed to absorb moisture
during the test. Resistivity of concrete decreases with an increase in temperature.7.19
The majority of values quoted in this section are given for alternating current (a.c).
The resistivity to direct current (d.c.) may be different because it has a polarizing

380
effect, but at 50 Hz there is no significant difference between resistivity to a.c. and
d.c.7.74 In general, for concrete matured in air, the d.c. resistance is approximately equal
to the a.c. impedance.7.74 Hammond and Robson7.74 interpreted this to mean that the
capacitative reactance of concrete is so much larger than its resistance that it is only
the latter that contributes significantly to impedance; as a consequence, the power
factor is nearly unity. Typical data for alternating current are given in Table 7.4.
The capacitance of concrete decreases with age and with an increase in
frequency.7.74 Neat cement paste with a water/cement ratio of 0.23 has a much higher
capacitance than concrete with a water/cement ratio of 0.49 at the same age.7.74
Data on the dielectric strength of concrete are given in Table 7.5. It can be seen
that the dielectric strength of concrete made with high-alumina cement is slightly
greater than when Portland cement is used. The table shows also that, despite the
much higher moisture content (and therefore lower resistivity) of air-stored concrete
compared with oven-dried concrete, the dielectric strength is approximately the same
for the two storage conditions, and seems thus to be unaffected by moisture content.

Table 7.5. Dielectric Strength of Concrete (1:2:4 Mix with Water/Cement Ratio of
0.49)7.74

Acoustic properties
In many buildings, acoustic characteristics are of importance and these may be greatly
influenced by the material used and by structural details. Here, only the properties of
the material will be considered, the influence of the structural form and construction
details being a specialized topic.
Basically, two acoustic properties of a building material can be distinguished:
sound absorption and sound transmission. The former is of interest when the source of
sound and the listener are in the same room. Energy of sound waves, when they hit a
wall, is partly absorbed and partly reflected, and we can define a sound absorption
coefficient as a measure of the proportion of the sound energy striking a surface which

381
is absorbed by that surface. The coefficient is usually given for a particular frequency.
Sometimes, the term ‘noise reduction coefficient’ is used to denote the average of
sound absorption coefficients at 250, 500, 1000, and 2000 Hz in octave steps. A
typical value for normal weight aggregate concrete of medium texture, unpainted, is
0.27. The corresponding value for concrete made with expanded shale aggregate is
0.45. The difference is related to the variation in texture, porosity and structure
because, when airflow is possible, there is a large increase in sound absorption
through conversion of sound energy into heat, by friction. Thus, cellular concrete,
which has discrete air bubbles, exhibits lower sound absorption than concrete made
with porous lightweight aggregate.
Sound transmission is of interest when the listener is in a room adjacent to that in
which the source of sound is located. We define the sound transmission loss (or
airborne-sound insulation) as the difference, measured in decibels (dB), between the
incident sound energy and the transmitted sound energy (which radiates into an
adjoining room). What constitutes a satisfactory transmission loss depends on the use
of the given space: a value of 45 to 55 dB is thought to be adequate between
dwellings.7.22,7.25
The primary factor in transmission loss is the unit mass of the partition per square
metre of area. The loss increases with the frequency of the sound wave and is usually
assessed over a range of frequencies. The relation between the transmission loss and
the mass of the partition, in general terms, is independent of the type of material used,
provided no continuous pores are present, and is sometimes referred to as the ‘mass
law’. Figure 7.27 illustrates the relation for the case when the partition edges are
‘firmly fixed’, i.e. the flanking walls are of similar material. From Fig. 7.27, it can be
seen that a bare concrete wall 150 to 175 mm (6 to 7 in.) thick would provide an
adequate transmission loss between dwellings. Information on sound insulation of
party walls is given in refs 7.22, 7.23 and 7.24; more general treatment of acoustic
properties of concrete is given in ref. 7.26.

382
 Fig. 7.27. Relation between transmission loss and unit mass of partition 7.68
The sound transmission around the ‘sound obstacle’ has, of course, to be
considered but, as far as the partition itself is concerned, there are some factors
additional to the mass: airtightness, bending stiffness, and the presence of cavities.
The stiffness of the partition is relevant because, if the wavelength of the forced
bending wave imposed on the wall is equal to the wavelength of free bending waves
in a wall, a condition of total sound transmission through the wall arises. This
coincidence of wavelengths can occur only above a critical value of frequency at
which the velocity of free bending waves in the wall is the same as that of air waves
parallel to the wall. Above that frequency, a combination of air-wave incidence and
frequency is possible at which there can occur the coincidence of air wave at interface
and of the structure bending wave. The effect is usually limited to thin walls.7.68 The
critical frequency is given by:

where v = velocity of sound in air

h = thickness of the partition
ρ = density of the concrete
E = modulus of elasticity of concrete, and
μ = Poisson’s ratio of concrete.

383
 The influence of the coincidence effect on the relation between the sound
transmission loss and the unit mass of the partition can be seen from the dotted line
in Fig. 7.27.
The presence of cavities also affects this relation, a cavity increasing the
transmission loss, so that the use of the given total thickness of concrete in the form of
two leaves is advantageous. The quantitative behaviour depends on the width of the
cavity, on the degree of isolation between the leaves, and also on the presence or
absence of a sealed surface facing the cavity if the wall material is porous.
From the foregoing it is apparent that, to a considerable extent, the requirements of
a high sound absorption and a high transmission loss are conflicting. For instance, the
porous type of lightweight concrete has good sound-absorbing properties but a very
high sound transmission. However, if one concrete face is sealed, the transmission
loss is increased and can become equal to that of other materials of the same mass per
unit area. It is preferable to seal the side remote from the source of the sound as,
otherwise, sound absorption is impaired. However, there is no reason to believe that
lightweight concrete provides inherently better insulation with respect to sound
transmission.
References
7.1. G. W. WASHA, J. C. SAEMANN and S. M. CRAMER Fifty-year properties of
concrete made in 1937, ACI Materials Journal, 86, No. 4, pp. 367–71 (1989).
7.2. L. J. PARROTT, Moisture profiles in drying concrete, Advances in Cement
Research, 1, No. 3, pp. 164–70 (1988).
7.3. R. G. PATEL, D. C. KILLOH, L. J. PARROTT and W. A. GUTTERIDGE, Influence of
curing at different relative humidities upon compound reactions and porosity
of Portland cement paste, Materials and Structures, 21, No. 123, pp. 192–7
(1988).
7.4. E. SENBETTA, Concrete curing practices in the United States, Concrete
International, 10, No. 11, pp. 64–7 (1988).
7.5. D. W. S. HO, Q. Y. CUI and D. J. RITCHIE, Influence of humidity and curing
time on the quality of concrete, Cement and Concrete Research, 19, No. 3,
pp. 457–64 (1989).
7.6. B. MATHER, Curing compounds, Concrete International, 12, No. 2, pp. 40–1
(1990).
7.7. P. NISCHER, General report: effects of early overloading and insufficient
curing on the properties of concrete after complete hardening,
in Proceedings of RILEM International Conference on Concrete of Early
Ages, Vol. II, pp. 117–26 (Paris, Anciens ENPC, 1982).
7.8. T. A. HARRISON, Formwork Striking Times – Methods of Assessment, Report
73, 40 pp. (London, CIRIA, 1987).
7.9. ACI 308-92, Standard practice for curing concrete, ACI Manual of Concrete
Practice, Part 2: Construction Practices and Inspection Pavements, 11 pp.
(Detroit, Michigan, 1994).
7.10. H. F. GONNERMAN and W. LERCH, Changes in characteristics of portland
cement as exhibited by laboratory tests over the period 1904 to 1950, ASTM
Sp. Tech. Publ. No. 127 (1951).

384
 7.11. W. H. PRICE, Factors influencing concrete strength, J. Amer. Concr. Inst., 47,
pp. 417–32 (Feb. 1951).
7.12. T. S. POOLE, Summary of statistical analyses of specification mortar cube test
results from various cement suppliers, including four types of cement
approved for Corps of Engineers projects, in Uniformity of Cement Strength
ASTM Sp. Tech. Publ. No. 961, pp. 14–21 (Philadelphia, Pa, 1986).
7.13. J. R. OGLESBY, Experience with cement strength uniformity, in Uniformity of
Cement Strength ASTM Sp. Tech. Publ. No. 961, pp. 3–14 (Philadelphia, Pa,
1986).
7.14. R. D. GAYNOR, Cement Strength Data for 1991, ASTM Committee C-1 on
Cement, 4 pp. (Philadelphia, Pa, 1993).
7.15. L. DIVET, Évolution de la composition des ciments Portland artificiels de
1964 à 1989: Exemple d’utilization de la banque de données du LCPC sur
les ciments, Bulletin Liaison Laboratoire Ponts et Chaussées, 176, pp. 73–
80 (Nov.–Dec. 1991).
7.16. A. T. CORISH and P. J. JACKSON, Portland cement properties, Concrete, 16, No.
7, pp. 16–18 (1982).
7.17. A. M. NEVILLE, Why we have concrete durability problems, in Concrete
Durability: Katharine and Bryant Mather International Conference, Vol. 1,
ACI SP-100, pp. 21–30 (Detroit, Michigan, 1987).
7.18. B. P. HUGHES, A. K. O. SOLEIT and R. W. BRIERLEY, New technique for
determining the electrical resistivity of concrete, Mag. Concr. Res., 37, No.
133, pp. 243–8 (1985).
7.19. H. W. WHITTINGTON, J. MCCARTER and M. C. FORDE, The conduction of
electricity through concrete, Mag. Concr. Res., 33, No. 114, pp. 48–60
(1981).
7.20. P. J. NIXON, Changes in Portland Cement Properties and their Effects on
Concrete, Building Research Establishment Information Paper, 3 pp. (March
1986).
7.21. CONCRETE SOCIETY WORKING PARTY, Report on Changes in Cement Properties
and their Effects on Concrete, Technical Report No. 29, 15 pp. (Slough,
U.K., 1987).
7.22. A. LITVIN and H. B. BELLISTON, Sound transmission loss through concrete and
concrete masonry walls, J. Amer. Concr. Inst., 75, pp. 641–6 (Dec. 1978).
7.23. BUILDING RESEARCH ESTABLISHMENT Sound Insulation in Party Walls, Digest No.
252, 4 pp. (Aug. 1981).
7.24. BUILDING RESEARCH ESTABLISHMENT Sound Insulation: Basic Principles, Digest
No. 337, 8 pp. (Oct. 1988).
7.25. A. C. C. WARNOCK, Factors Affecting Sound Transmission Loss, Canadian
Building Digest, CDN 239, 4 pp. (July 1985).
7.26. C. HUET, Propriétés acoustiques in Le béton hydraulique, pp. 423–52 (Paris,
Presses de l’École Nationale des Ponts et Chaussées, 1982).
7.27. U. B. HALABE, A. SOTOODEHNIA, K. R. MASER and E. A. KAUSEL, Modeling the
electromagnetic properties of concrete, ACI Materials Journal, 90, No. 6, pp.
552–63 (1993).

385
 7.28. P. SCHIESSL and C. REUTER, Massgebende Einflussgrössen auf die Wasser-
durchlässigkeit von gerissenen Stahlbetonbauteilen, Annual Report, Institut
für Bauforschung, Aachen, pp. 223–8 (1992).
7.29. M. BEN-BASSAT, P. J. NIXON and J. HARDCASTLE, The effect of differences in the
composition of Portland cement on the properties of hardened concrete, Mag.
Conc. Res., 42, No. 151, pp. 59–66 (1990).
7.30. I. L. H. HANSSON and C. M. HANSSON, Electrical resistivity measurements of
portland cement based material, Cement and Concrete Research, 13, No. 5,
pp. 675–83 (1983).
7.31. Y. ABDEL-JAWAD and R. HADDAD, Effect of early overloading of concrete on
strength at later ages, Cement and Concrete Research, 22, No. 5, pp. 927–36
(1992).
7.32. W. S. WEAVER, H. L. ISABELLE and F. WILLIAMSON, A study of cement and
concrete correlation, Journal of Testing and Evaluation, 2, No. 4, pp. 260–
303 (1974).
7.33. A. NEVILLE, Cement and concrete: their interaction in practice, in Advances in
Cement and Concrete, American Soc. Civil Engineers, pp. 1–14 (New York,
1994).
7.34. J. TRITTHART and H. G. GEYMAYER, Änderungen des elektrischen Widerstandes
in austrocknendem Beton, Zement und Beton, 30, No. 1, pp. 23–8 (1985).
7.35. L. E. COPELAND and R. H. BRAGG, Self-desiccation in portland cement
pastes, ASTM Bull. No. 204, pp. 34–9 (Feb. 1955).
7.36. T. C. POWERS, A discussion of cement hydration in relation to the curing of
concrete, Proc. Highw. Res. Bd, 27, pp. 178–88 (Washington DC, 1947).
7.37. W. LERCH, Plastic shrinkage, J. Amer. Concr. Inst., 53, pp. 797–802 (Feb.
1957).
7.38. A. D. Ross, Shape, size, and shrinkage, Concrete and Constructional
Engineering, pp. 193–9 (London, Aug. 1944).
7.39. F. R. MCMILLAN and L. H. TUTHILL, Concete primer, ACI SP-1 3rd Edn, 96 pp.
(Detroit, Michigan, 1973).
7.40. U.S. ARMY CORPS OF ENGINEERS, Handbook for Concrete and
Cement (Vicksburg, Miss., 1954).
7.41. F. M. LEA, Would the strength grading of ordinary Portland cement be a
contribution to structural economy? Proc. Inst. Civ. Engrs, 2, No. 3, pp.
450–7 (London, Dec. 1953).
7.42. S. WALKER and D. L. BLOEM, Variations in portland cement, Proc. ASTM, 58,
pp. 1009–32 (1958).
7.43. E. W. BENNETT and N. K. RAJU, Cumulative fatigue damage of plain concrete
in compression, Proc. Int. Conf. on Structure, Solid Mechanics and
Engineering Design, Southampton, April 1969, Part 2, pp. 1089–102 (New
York, Wiley-Interscience, 1971).
7.44. J. P. LLOYD, J. L. LOTT and C. E. KESLER, Final summary report: fatigue of
concrete, T. & A. M. Report No. 675, Department of Theoretical and
Applied Mechanics, University of Illinois, 33 pp. (Sept. 1967).

386
 7.45. A. M. NEVILLE, Current problems regarding concrete under sustained
loading, Int. Assoc. for Bridge and Structural Engineering, Publications, No.
26, pp. 337–43 (1966).
7.46. F. S. OPLE JR and C. L. HULSBOS, Probable fatigue life of plain concrete with
stress gradient, J. Amer. Concr. Inst., 63, pp. 59–81 (Jan. 1966).
7.47. J. W. MURDOCK, The mechanism of fatigue failure in concrete, Thesis
submitted to the University of Illinois for the degree of Ph.D., 131 pp.
(1960).
7.48. J. A. NEAL and C. E. KESLER, The fatigue of plain concrete, Proc. Int. Conf.
on the Structure of Concrete, pp. 226–37 (London, Cement and Concrete
Assoc., 1968).
7.49. B. M. ASSIMACOPOULOS, R. F. WARNER and C. E. EKBERG, JR, High speed fatigue
tests on small specimens of plain concrete, J. Prestressed Concr. Inst., 4, pp.
53–70 (Sept. 1959).
7.50. W. H. GRAY, J. F. MCLAUGHLIN and J. D. ANTRIM, Fatigue properties of
lightweight aggregate concrete, J. Amer. Concr. Inst., 58, pp. 149–62 (Aug.
1961).
7.51. A. M. OZELL, Discussion of paper by J. P. ROMUALDI and G. B.
BATSON: Mechanics of crack arrest in concrete, J. Eng. Mech.
Div., A.S.C.E., 89, No. EM 4, p. 103 (Aug. 1963).
7.52. H. GREEN, Impact strength of concrete, Proc. Inst. Civ. Engrs., 28, pp. 383–
96 (London, July 1964).
7.53. G. B. WELCH and B. HAISMAN, Fracture toughness measurements of
concrete, Report No. R42, University of New South Wales, Kensington,
Australia (Jan. 1969).
7.54. B. P. HUGHES and R. GREGORY, The impact strength of concrete using Green’s
ballistic pendulum, Proc. Inst. Civ. Engrs., 41, pp. 731–50 (London, Dec.
1968).
7.55. U. BELLANDER, Concrete strength in finished structure, Part 1: Destructive
testing methods. Reasonable requirements, CBI Research, 13: 76, 205 pp.
(Swedish Cement and Concrete Research Inst., 1976).
7.56. D. C. TEYCHENNÉ, Concrete made with crushed rock aggregates, Quarry
Management and Products, 5, pp. 122–37 (May 1978).
7.57. R. L. MCKISSON, Cement uniformity on Bureau of Reclamation projects, U.S.
Bureau of Reclamation, Laboratory Report C-1245, 41 pp. (Denver,
Colorado, Aug. 1967).
7.58. S. S. TAKHAR, I. J. JORDAAN and B. R. GAMBLE, Fatigue of concrete under lateral
confining pressure, in Abeles Symp. on Fatigue of Concrete, ACI SP-41, pp.
59–69 (Detroit, Michigan, 1974).
7.59. K. D. RAITHBY and J. W. GALLOWAY, Effects of moisture condition, age, and
rate of loading on fatigue of plain concrete, in Abeles Symp. on Fatigue of
Concrete, ACI SP-41, pp. 15–34 (Detroit, Michigan, 1974).
7.60. H. SOMMER, Zum Einfluss der Kornzusammensetzung auf die Dauerfestigkeit
von Beton, Zement und Beton, 22, No. 3, pp. 106–9 (1977).

387
 7.61. R. TEPFERS and T. KUTTI, Fatigue strength of plain, ordinary and lightweight
concrete, J. Amer. Concr. Inst., 76, No. 5, pp. 635–52 (1979).
7.62. R. TEPFERS, C. FRIDÉN and L. GEORGSSON, A study of the applicability to the
fatigue of concrete of the Palmgren–Miner partial damage hypothesis, Mag.
Concr. Res., 29, No. 100, pp. 123–30 (1977).
7.63. R. TEPFERS, Tensile fatigue strength of plain concrete, J. Amer. Concr.
Inst., 76, No. 8, pp. 919–33 (1979).
7.64. J. W. GALLOWAY, H. M. HARDING and K. D. RAITHBY, Effects of age on flexural,
fatigue and compressive strength of concrete, Transport and Road Res. Lab.
Rep. TRRL 865, 20 pp. (Crowthorne, Berks., 1979).
7.65. J. VAN LEEUWEN and A. J. M. SIEMES, Miner’s rule with respect to plain
concrete, Heron, 24, No. 1, 34 pp. (Delft, 1979).
7.66. J. DAHMS, Die Schlagfestigkeit des Betons, Schriftenreihe der Zement
Industrie, No. 34, 135 pp. (Düsseldorf, 1968).
7.67. C. POPP, Untersuchen über das Verhalten von Beton bei schlagartigen
Beanspruchung, Deutscher Ausschuss für Stahlbeton, No. 281, 66 pp.
(Berlin, 1977).
7.68. A. G. LOUDON and E. F. STACEY, The thermal and acoustic properties of
lightweight concretes, Structural Concrete, 3, No. 2, pp. 58–96 (London,
1966).
7.69. D. A. HAUSMANN, Electrochemical behavior of steel in concrete, J. Amer.
Concr. Inst., 61, No. 2, pp. 171–88 (Feb. 1964).
7.70. G. E. MONFORE, The electrical resistivity of concrete, J. Portl. Cem. Assoc.
Research and Development Laboratories, 10, No. 2, pp. 35–48 (May 1968).
7.71. R. CIGNA, Measurement of the electrical conductivity of cement
mortars, Annali di Chimica, 66, pp. 483–94 (Jan. 1966).
7.72. R. L. HENRY, Water vapor transmission and electrical resistivity of
concrete, Technical Report R-244 (US Naval Civil Engineering Laboratory,
Port Hueneme, California, 30 June 1963).
7.73. V. P. GANIN, Electrical resistance of concrete as a function of its
composition, Beton i Zhelezobeton, No. 10, pp. 462–5 (1964).
7.74. E. HAMMOND and T. D. ROBSON, Comparison of electrical properties of various
cements and concretes, The Engineer, 199, pp. 78–80 (21 Jan. 1955); pp.
114–15 (28 Jan. 1955).
7.75. ANON, Electrical properties of concrete, Concrete and Constructional
Engineering, 58, No. 5, p. 195 (London, 1963).
7.76. J. R. FARRAR, Electrically conductive concrete, GEC J. of Science and
Technol., 45, No. 1, pp. 45–8 (1978).
7.77. P. KLIEGER, Early high strength concrete for prestressing, Proc. of World
Conference on Prestressed Concrete, pp. A5-1–14 (San Francisco, July
1957).
7.78. B. M. SCOTT, Cement strength uniformity – a ready-mix producer’s point of
view, NRMCA Publication No. 165, 3 pp. (Silver Spring, Maryland, 1981).
7.79. P. ROSSI et al. Effect of loading rate on the strength of concrete subjected to
uniaxial tension, Materials and Structures, 27, No. 169, pp. 260–4 (1994).

388
Chapter 8. Temperature effects in concrete

Laboratory testing of concrete is usually performed at a controlled temperature,

normally constant. As the early testing was done in temperate climates, the
standardized temperature chosen was generally in the region of 18 to 21 °C (64 to
70 °F) so that much of the basic information about the properties of both fresh and
hardened concrete is based on the behaviour of concrete at these temperatures. In
practice, however, concrete is mixed at a wide range of temperatures and also remains
in service at different temperatures. Indeed, the actual range of temperatures has
widened considerably with much modern construction taking place in countries which
have a hot climate. Also, new developments, mainly offshore, take place in very cold
regions.
In consequence, knowledge of the temperature effects in concrete is of great
importance. These effects will be considered in the present chapter. First, the
influence of the temperature of fresh concrete upon strength will be discussed; this
will be followed by a review of temperature treatment after the placing of concrete,
that is curing both by using steam at atmospheric pressure and by high-pressure steam.
Next, the effects of the temperature rise in concrete due to the development of the heat
of hydration of cement will be discussed, followed by consideration of concreting in
hot weather and in cold weather. Finally, thermal properties of hardened concrete will
be described and the influence of very high and very low temperatures in service,
including the effects of fire, will be discussed.
Influence of early temperature on strength of concrete
We have seen that a rise in the curing temperature speeds up the chemical reactions of
hydration and thus affects beneficially the early strength of concrete without any ill-
effects on the later strength. Higher temperature during and following the initial
contact between cement and water reduces the length of the dormant period so that
the overall structure of the hydrated cement paste becomes established very early.
Although a higher temperature during placing and setting increases the very early
strength, it may adversely affect the strength from about 7 days onwards. The
explanation is that a rapid initial hydration appears to form products of a poorer
physical structure, probably more porous, so that a proportion of the pores will always
remain unfilled. It follows from the gel/space ratio rule, that this will lead to a lower
strength compared with a less porous, though slowly hydrating, cement paste in which
a high gel/space ratio will eventually be reached.
This explanation of the adverse effects of a high early temperature on later strength
has been extended by Verbeck and Helmuth8.77 who suggest that the rapid initial rate of
hydration at higher temperatures retards the subsequent hydration and produced a
non-uniform distribution of the products of hydration within the paste. The reason for
this is that, at the high initial rate of hydration, there is insufficient time available for
the diffusion of the products of hydration away from the cement particle and for a
uniform precipitation in the interstitial space (as is the case at lower temperatures). As
a result, a high concentration of the products of hydration is built up in the vicinity of
the hydrating particles, and this retards the subsequent hydration and adversely affects
the long-term strength. The presence of porous C-S-H in between the cement particles
has been confirmed by backscattered electron imaging.8.74

389
 In addition, the non-uniform distribution of the products of hydration per
se adversely affects the strength because the gel/space ratio in the interstices is lower
than would be otherwise the case for an equal degree of hydration: the local weaker
areas lower the strength of the hydrated cement paste as a whole.
In connection with the influence of temperature during the early life of concrete on
the overall structure of the hydrated cement paste, it is useful to recall that a low early
gain of strength has a beneficial effect on strength also when the hydration is slowed
down by the use of retarders. Water-reducing and set-retarding admixtures were found
to be beneficial in compensating for the reduction in the long-term strength of
admixture-free concrete placed at a high temperature.8.24 It should be realized, however,
that their effect arises from water reduction and therefore a lower water/cement
ratio.8.14 Moreover, the rate of loss of slump is higher when these admixtures are
used.8.14
Figure 8.1 shows Price’s8.11 data on the effect of the temperature during the first two
hours after mixing on the development of strength of concrete with a water/cement
ratio of 0.53. The range of temperatures investigated was 4 to 46 °C (40 to 115 °F)
and, beyond the age of two hours, all specimens were cured at 21 °C (70 °F). The
specimens were sealed so as to prevent movement of moisture. Tests on cylinders
moist-cured during the first 24 hours at 2 °C (36 °F) and at 18 °C (64 °F), and
thereafter at 18 °C (64 °F) have shown that, at 28 days, the former are 10 per cent
stronger than the latter.8.80

390
 Fig. 8.1. Effect of temperature during the first two hours after casting on the
development of strength (all specimens sealed and after the age of 2 hours cured
at 21 °C (70 °F))8.11
Some other test data are given below but direct comparisons are difficult because
varying combinations of temperature and time were used in the various studies. An
increase in the 24-hour strength of concrete, coupled with a decrease in the 28-day
strength, in consequence of a higher temperature during the first 4 hours, was
observed by Petscharnig8.26 (see Fig. 8.2). He found the effect to be more pronounced
with a more rapid-hardening cement and with a higher cement content.

Fig. 8.2. Influence of initial temperature on the average monthly compressive

strength of concrete cured at a constant temperature from the age of four hours
onwards: the temperature can be inferred from the time of the year when the
concrete specimens were made in the open in Austria (based on ref. 8.26)
Temperature of 38 °C (100 °F) during the first 24 hours was reported to result in a
loss of strength of concrete at 28 days of 9 to 12 per cent, compared with the same
concrete cured at 23 °C (70 °F) throughout;8.25 the concrete had a 28-day standard
cylinder strength of 28 MPa (4000 psi).
A review of the effect of a higher temperature during the first few days on the
strength of test cylinders,8.58 as compared with cylinders cured in a standard manner,
has shown a significant reduction in the recorded 28-day strength: 1 day at 38 °C
(100 °F) results in a reduction of about 10 per cent, and 3 days at 38 °C in a reduction
of about 22 per cent.
Some field tests have confirmed the influence of temperature at the time of placing
on strength: typically, for an increase of 5 °C (9 °F) there is a decrease in strength of
1.9 MPa (270 psi).8.85
The influence of the temperature in the early life of cement paste (from the age of
24 hours onwards) on the structure of the hydrated cement paste was demonstrated by
Goto and Roy8.113 who found that curing at 60 °C (140 °F) results in a much higher
volume of pores larger than 150 nm in diameter, compared with curing at 27 °C

391
(81 °F). The total porosity varied in the opposite direction, but it is the larger pores
that control permeability, which is of great importance with regard to durability.
The influence of curing temperature on strength of concrete (tested after cooling)
at 1 and 28 days is shown in Fig. 8.3.8.77 However, the temperature at the time of
testing also appears to be a factor, at least in the case of neat (ordinary Portland)
cement paste compacts with a water/cement ratio of 0.14.8.81 The temperature was kept
constant from the initiation of hydration. When tested (at 64 and 128 days) at the
curing temperature, the specimens had a lower strength at higher temperatures (Fig.
8.4); but, if cooled to 20 °C (68 °F) over a period of two hours prior to testing, only
temperatures above 65 °C (150 °F) had a deleterious effect (Fig. 8.5).

Fig. 8.3. Influence of curing temperature on compressive strength at 1 and 28

days (specimens tested after cooling to 23 °C (73 °F) over a period of two
hours)8.77

392
 Fig. 8.4. Relation between compressive strength and curing time of neat cement
paste compacts at different curing temperatures. The temperature of the
specimens was kept constant up to and including the period of testing8.81

393
 Fig. 8.5. Relation between compressive strength and curing time of neat cement
paste compacts at different curing temperatures. The temperature of the
specimens was moderated to 20 °C at a constant rate over a two-hour period
prior to testing (water/cement ratio = 0.14; Type I cement)8.81
Tests have also been made on concretes stored in water at different temperatures
for a period of 28 days, and thereafter at 23 °C (73 °F).8.70 As in Price’s tests, a higher
temperature was found to result in a higher strength during the first few days after
casting but, beyond the age of one to four weeks, the situation changed radically. The

394
specimens cured at temperatures between 4 and 23 °C (40 and 73 °F) up to the age of
28 days all showed a higher strength than those cured at 32 to 49 °C (90 to 120 °F).
Among the latter, retrogression was greater the higher the temperature but, in the
lower range of temperatures, there appeared to be an optimum temperature that
yielded the highest strength. It is interesting to note that even concrete cast at 4 °C
(40 °F) and stored at as low a temperature as –4 °C (25 °F) for four weeks and then at
23 °C (73 °F) is from the age of 3 months onwards stronger than similar concrete
stored continuously at 23 °C (73 °F). Figure 8.6 shows typical curves for concrete
containing 307 kg of ordinary Portland cement per cubic metre of concrete (517 lb/yd3)
with 4.5 per cent of entrained air. Similar behaviour has been observed when rapid-
hardening Portland and modified cement are used.

Fig. 8.6. Effect of temperature during the first 28 days on the strength of
concrete (water/cement ratio = 0.41; air content = 4.5 per cent; ordinary
Portland cement)8.70
In concrete members with a high cement content, as is the case with high
performance concrete, there is a considerable temperature rise even in ordinary
structural elements such as beams and columns. The 7-day strength is higher the
greater the temperature rise; for example, when the temperature was 20 °C (68 °F) the
strength was 96 MPa, but when the maximum temperature was 75 °C (167 °F) the
strength was 115 MPa. However, at 28 days, there was a reversal in the strength
values with the low temperature leading to a strength of 122 MPa while the high
temperature led to a reduced strength of 112 MPa. Maximum temperatures between
45 and 65 °C led to a very slight increase in strength between the ages of 7 and 28
days.8.57

395
 With respect to the strength of concrete cured at very low temperatures, Aïtcin et
al. found that, provided concrete with a water/cement ratio between 0.45 and 0.55
8.23

was cast and maintained for 9 hours at a temperature not lower than 4 °C (39 °F),
subsequent storage in sea water at 0 °C (32 °F) led to an increase in strength. The
increase was at first very slow but at the age of 4 days the specimens immersed in sea
water reached about one-half of the strength of standard-cured specimens. The
difference between the strengths for the two storage conditions gradually decreased,
after 2 months becoming about 10 MPa (1500 psi); this value persisted for at least a
year. Concrete with a lower water/cement ratio performed better than concretes with
higher water/cement ratios.8.18,8.23
Klieger’s tests8.70 indicate that there is an optimum temperature during the early life
of concrete that will lead to the highest strength at a desired age. For laboratory-made
concrete, using ordinary or modified Portland cement, the optimum temperature is
approximately 13 °C (55 °F); for rapid-hardening Portland cement it is about 4 °C
(40 °F). It must not be forgotten, however, that beyond the initial period of setting and
hardening the influence of temperature (within limits) accords with the maturity rule:
a higher temperature accelerates the development of strength.
The tests described so far were all made in the laboratory or under known
conditions, but the behaviour on site in a hot climate may not be the same. Here, there
are some additional factors acting: ambient humidity, direct radiation of the sun, wind
velocity, and method of curing. It should be remembered also that the quality of
concrete depends on its temperature and not on that of the surrounding atmosphere, so
that the size of the member also enters the picture because it affects the rise in
temperature caused by the hydration of cement. Likewise, curing by flooding in
windy weather results in a loss of heat due to evaporation so that the temperature of
concrete is lower than when a sealing compound is used. These factors are discussed
later in the present chapter.
Steam curing at atmospheric pressure
Because an increase in the curing temperature of concrete increases its rate of
development of strength, the gain of strength can be speeded up by curing concrete in
steam. When steam is at atmospheric pressure, i.e. the temperature is below 100 °C
(212 °F), the process can be regarded as a special case of moist curing in which the
vapour-saturated atmosphere ensures a supply of water. In addition, condensation of
the steam releases latent heat. High-pressure steam curing (autoclaving) is an entirely
different operation and is considered in the next section.
The primary object of steam curing is to obtain a sufficiently high early strength so
that the concrete products may be handled soon after casting: the moulds can be
removed, or the prestressing bed vacated, earlier than would be the case with ordinary
moist curing, and less curing storage space is required; all these mean an economic
advantage. For many applications, the long-term strength of concrete is of lesser
importance.
Because of the nature of the operations involved in steam curing, the process is
used mainly with precast products. Low-pressure steam curing is normally applied in
special chambers or in tunnels through which the concrete members are transported
on a conveyor belt. Alternatively, portable boxes or plastic covers can be placed over
precast members, steam being supplied through flexible pipes.

396
 Due to the influence of temperature during the early stages of hardening on the
later strength, a compromise between the temperatures giving a high-early and a high-
late strength has to be made. Figure 8.7 shows typical values of strength of concrete
made with modified (Type II) cement and a water/cement ratio of 0.55; steam curing
was applied immediately after casting. Long-term retrogression of strength was
observed.

Fig. 8.7. Strength of concrete cured in steam at different temperatures

(water/cement ratio = 0.55; steam curing applied immediately after casting)8.71
A probable, possibly partial, explanation of the reduction in the long-term strength
of steam-cured concrete lies in the presence of very fine cracks caused by the
expansion of air bubbles in the cement paste; the thermal expansion of the air is at
least two orders of magnitude greater than that of the surrounding solid material. The
expansion of air bubbles is restrained so that the air is put under pressure and, to
balance this pressure, tensile stresses are induced in the surrounding cement paste.
These stresses may induce very fine cracks. Strictly speaking, therefore, we are
dealing not with a long-term loss of strength but with a loss at all ages.8.82 However, up
to the age of 28 days, this loss is masked by the beneficial effect on strength of a
higher temperature during curing.

397
 The role of the expanding air bubbles, as well as of water, is indirectly
demonstrated by the very high coefficient of thermal expansion of fresh concrete (30
× 10–6) compared with the coefficient after 4 hours (11.5 × 10–6) reported by
Mamillan.8.37
The disruptive effects of the expansion of air bubbles can be reduced by a
prolonged delay period prior to steam curing (during which the tensile strength of
concrete increases) or by a lower rate of temperature rise (as the increase in the air
pressure is matched by an increase in the strength of the surrounding cement paste).
Alternatively, heating in closed formwork or in pressure chambers can be
used.8.82 With short-term curing periods (2 to 5 hours) and moderate temperatures,
there is probably little real retrogression of strength, and the apparent low strength at
later ages is due to the absence of prolonged wet curing.8.83
Because the adverse effect of steam curing on the long-term strength of concrete
arises from the changes in porosity and pore size in hydrated cement paste, it can be
expected that steam curing affects the durability of concrete; this is discussed on
p. 485.
To minimize the long-term retrogression of strength, two aspects of a steam-curing
cycle should be controlled: the delay in the commencement of heating and the rate of
temperature rise.
Because it is the temperature at the time of setting that has the greatest influence
on the strength at later ages, a delay in the application of steam curing is advantageous.
Some indication of the influence of the delay in heating on strength can be obtained
from Fig. 8.8 plotted by Saul8.72 from the data of Shideler and Chamberlin.8.73 The
concrete used was made with Type II cement, and had a water/cement ratio of 0.6.
The solid line shows the gain in strength of moist-cured concrete at room temperature
plotted against maturity. The dotted lines refer to different curing temperatures
between 38 and 85 °C (100 and 185 °F), and the figure against each point denotes the
delay in hours before the higher curing temperature was suddenly applied.

398
 Fig. 8.8. Effect of delay in steam curing on the early gain of strength with
maturity.8.72 Small figures indicate the delay in hours before curing at the
temperature indicated
From Fig. 8.8 it can be seen, for each curing temperature, there is a part of the
curve showing a normal rate of gain in strength with maturity. In other words, after a
sufficient delay, rapid heating has no adverse effect. This delay is approximately 2, 3,
5, and 6 hours, respectively, for 38, 54, 74, and 85 °C (100, 130, 165, and 185 °F). If,
however, concrete is exposed to the higher temperature with a smaller delay, the
strength is adversely affected, as shown by the right-hand portion of each dotted curve;
this effect is more serious the higher the curing temperature. Without a delay period,
the loss in the 28-day strength of concrete with a water/cement ratio of 0.50, steam
cured at 75 °C (167 °F), can be as much as 40 per cent.8.37
An additional argument supporting the need for the delay period is that it allows
gypsum to react with C3A. At higher temperatures, the solubility of gypsum is
decreased so that some of it might not react with C3A and do so only later, causing an
expansive reaction of the type known as sulfate attack (see p. 508).8.31 This view has
not been confirmed.
Figure 8.8 shows also that, within a few hours of casting, the rate of gain in
strength is higher than would be expected from the maturity calculations. This
confirms the earlier observation that the age at which a higher temperature is applied
is a factor in the maturity rule.
The desirable length of the delay period (when the ambient temperature should
match that of the concrete) depends on the size and shape of the concrete elements

399
being steam cured, on the water content of the concrete, and on the type of cement:
when the rate of hardening is slow, the delay should be longer. However, if a large
surface area is exposed, fog spray may be required to prevent plastic shrinkage
cracking. Guidance on the choice of the delay period is given in ACI 517.2R-87
(Revised 1992).8.27
The subsequent rate of rise in temperature also has to be controlled, depending on
the nature of the concrete units, so as to prevent the development of steep temperature
gradients in the concrete. A trial-and-error approach is necessary. ACI 517.2R-87
(Revised 1992)8.27 recommends rates ranging between 33 °C (60 °F) per hour for small
units and 11 °C (20 °F) per hour for large units. The rate of temperature rise has little
effect on the long-term strength, but the maximum temperature is a factor: a
temperature of 70 to 80 °C (160 to 180 °F) results in a reduction in the 28-day
strength of about 5 per cent.8.27
This effect has to be balanced, in economic terms, by the fact that a lower
maximum temperature requires a longer steam-curing period. It should be noted,
however, that the supply of heat need not continue once the temperature of the
concrete has stabilized at the maximum value; this time interval is referred to as
‘soaking’.
The period of steam curing at the maximum temperature is followed by cooling.
This may be rapid in the case of small units, but in large units rapid cooling may lead
to surface cracking. Supplementary wet curing may be beneficial in preventing rapid
drying out and improving the subsequent increase in strength.8.83 Concrete with a lower
water/cement ratio responds to steam curing much better than a mix with a high
water/cement ratio.
In summary, a curing cycle consists of a delay period, known also as preset period,
a temperature-rise period, a steaming period (which includes soaking) at the
maximum temperature, and a cooling period, possibly followed by wet curing.
Practical curing cycles are chosen as a compromise between the early and late
strength requirements but are governed also by the time available (e.g. length of work
shifts). Economic considerations determine whether the curing cycle should be suited
to a given concrete mix or, alternatively, whether the mix ought to be chosen so as to
fit a convenient cycle of steam curing. Whereas details of an optimum curing cycle
depend on the type of concrete product treated, a typical satisfactory cycle would
consist of the following:8.27 a delay period of 2 to 5 hours, heating at the rate of 22 to
44 °C per hour (40 to 80 °F per hour) up to a maximum temperature of 50 to 82 °C
(122 to 180 °F), then storage at maximum temperature, and finally a cooling period,
the total cycle (exclusive of the delay period) occupying preferably not more than 18
hours.
For concrete which is to be exposed to aggressive conditions, CIRIA Report C660
published in 2007 gives guidance on the maximum temperature and on the rate of
temperature rise.
Lightweight aggregate concrete can be heated up to between 82 and 88 °C (180
and 190 °F), but the optimum cycle is no different from that for concrete made with
normal weight aggregate.8.79
Steam curing has been used successfully with different types of Portland cement,
as well as with blended cements, but must never be used with high-alumina cement
because of the adverse effect of hot, wet conditions on the strength of that cement.
Steam curing of concrete made with fly ash accelerates the pozzolanic reaction with

400
Ca(OH)2 but only above a temperature of 88 °C (190 °F). A similar situation obtains
with ground granulated blastfurnace slag in the mix above 60 °C (140 °F). An
increased fineness of the slag (above 600 m2/kg) is beneficial with respect to the
effects of steam curing on strength.8.28 The slag also leads to a reduction in the average
pore size in the steam-cured cement paste.8.28
High-pressure steam curing (autoclaving)
This process is quite different from curing in steam at atmospheric pressure, both in
the method of execution and in the nature of the resulting concrete.
Because pressures above atmospheric are involved, the curing chamber must be of
the pressure-vessel type with a supply of wet steam; superheated steam must not be
allowed to come into contact with the concrete because it would cause drying of
concrete. Such a vessel is known as an autoclave, and high-pressure steam curing is
also referred to as autoclaving.
High-pressure steam curing was first employed in the manufacture of sand–lime
brick and of lightweight cellular concrete, and is still extensively used for that purpose.
In the field of concrete, high-pressure steam curing is usually applied to precast
products, generally small, but also to bridge truss members (made both of normal
weight and lightweight concrete) when any of the following characteristics are desired:
(a) high early strength: with high-pressure steam curing, the 28-day strength on
normal curing can be reached in about 24 hours; strengths of 80 to 100 MPa
(12 000 to 15 000 psi) have been reported;8.29
(b) high durability: high-pressure steam curing improves the resistance of
concrete to sulfates and to other forms of chemical attack, also to freezing and
thawing, and reduces efflorescence; and
(c) reduced drying shrinkage and moisture movement.
The optimum curing temperature has been found experimentally to be about
177 °C (350 °F)8.75 which corresponds to a steam pressure of 0.8 MPa (120 psi) above
atmospheric pressure.
High-pressure steam curing is most effective when finely ground silica is added to
the cement, owing to the chemical reactions between the silica and Ca(OH)2 released
on hydration of C3S (see Fig. 8.9). Cements rich in C3S have a greater capacity for
developing high strength when cured at high pressure than those with a high C2S
content, although, for short periods of high-pressure steam curing, cements with a
moderately low C3S/C2S ratio give good results.8.76 The high temperature during curing
affects also the reactions of hydration of the cement itself. For instance, some of the
C3S may hydrate to C3SHx.

401
 Fig. 8.9. Influence of pulverized silica content on the strength of high-pressure
steam-cured concrete (age at commencement of curing, 24 hours; curing
temperature, 177 °C (350 °F))8.75
The fineness of the silica should be at least equal to that of the cement; a higher
fineness, 600 m2/kg, was found8.29 to lead to an increase in strength of 7 to 17 per cent
compared with silica having a fineness of 200 m2/kg. Cement and silica must be
intimately mixed before they are fed into the mixer. The optimum amount of silica
depends on the mix proportions but is generally between 0.4 and 0.7 of the mass of
cement.
It is essential that the rate of heating during high-pressure steam curing is not too
high, as interference with the setting and hardening processes may occur in a manner
similar to that discussed in connection with steam curing at atmospheric pressure. A
typical steaming cycle consists of a gradual increase to the maximum temperature of
182 °C (360 °F) (which corresponds to a pressure of 1 MPa (145 psi)) over a period of
3 hours. This is followed by 5 to 8 hours at this temperature, and then a release of
pressure in about 20 to 30 minutes. A rapid release accelerates the drying of the
concrete so that shrinkage in situ will be reduced. At each temperature there is an
optimum period of curing (see Fig. 8.10).8.84

402
Fig. 8.10. Strength development of concrete at different curing temperatures for
various periods of curing8.84
It is worth emphasizing that a longer period of curing at a lower temperature leads
to a higher optimum strength than when high temperature is applied for a shorter time.
For any one period of curing, there is a temperature which leads to an optimum
strength. Also, for a given set of materials, it is possible to draw a line joining the
points of optimum strength at various curing periods and the curing
temperature;8.84 this is shown in Fig. 8.10.
In practice, the details of the steaming cycle depend on the plant used and also on
the size of the concrete members being cured. The length of the period of normal
curing preceding placing in the autoclave does not affect the quality of the steam-
cured concrete, and the choice of a suitable period is governed by the stiffness of the
mix, which must be strong enough to withstand handling. In the case of lightweight
concretes, the details of the steaming cycle have to be determined experimentally so
suit the materials used.

403
 Steam curing should be applied to concretes made with Portland cement only:
high-alumina and supersulfated cements would be adversely affected by the high
temperature.
Within the Portland group, the type of cement affects the strength, but not
necessarily in the same way as at normal temperatures; no systematic studies have,
however, been made. It is known, though, that ground granulated blastfurnace slag
may cause trouble if it has a high sulfur content. High-pressure steam curing
accelerates the hardening of concrete containing calcium chloride, but the relative
increase in stength is less than when no calcium chloride is used.
High-pressure steam curing produces a hydrated cement paste of low specific
surface, about 7000 m2/kg. Because the specific surface of high-pressure steam-cured
paste is thus only about of that of cement cured at ordinary temperature, it appears
that no more than 5 per cent of the high-pressure cured paste can be classified as gel.
This means that the products of hydration are coarse and largely microcrystalline. For
this reason, high-pressure steam-cured concrete has a considerably reduced shrinkage,
about to of that of concrete cured at normal temperatures. When silica is added to
the mix, shrinkage is higher, but still only about one-half of the shrinkage of normally
cured concrete. By contrast, because low-pressure steam curing does not produce a
micro-crystalline hydrated cement paste, no reduction in shrinkage is obtained. Creep
is also significantly reduced by high-pressure steam curing.
The products of hydration of cement subjected to high-pressure steam curing, as
well as those of the secondary lime–silica reactions, are stable, and there is no
retrogression of strength. At the age of one year, the strength of normally cured
concrete is approximately the same as that of high-pressure steam-cured concrete of
similar mix proportions. The water/cement ratio affects the strength of high-pressure
steam-cured concrete in the usual manner, but the actual values of early strength differ,
of course, from those for ordinary curing. The coefficient of thermal expansion and
the modulus of elasticity of concrete seem unaffected by high-pressure steam
curing.8.75
High-pressure steam curing improves the resistance of concrete to sulfate attack.
This is due to several reasons, the main one being the formation of aluminates more
stable in the presence of sulfates than those formed at lower temperatures. For this
reason, the relative improvement in resistance to sulfate attack is greater in cements
with a high C3A content than in cements which are resistant to sulfates. Another
important factor is the reduction in lime in the cement paste as a result of the lime–
silica reaction. Further improvement in sulfate resistance is due to the increased
strength and lower permeability of the steam-cured concrete, and also to the existence
of hydrates in a well-crystallized form.
High-pressure steam curing reduces efflorescence as there is no lime left to be
leached out.
High-pressure steam-cured concrete tends to be rather brittle. High-pressure steam
curing may reduce the strength in bond with plain reinforcement but not with
deformed bars. Good impact strength of high-pressure steam-cured concrete has been
reported.8.86 On the whole, high-pressure steam curing produces good quality, dense
and durable concrete. It is whitish in appearance as distinct from the characteristic
colour of normally-cured Portland cement concrete.
Other thermal curing methods

404
There exist several other methods of applying heat to concrete for the purpose of
accelerating the gain of strength. They are all specialized and applicable only in
certain cases. For this reason, no more than a brief mention will be given below.
The hot-mix method relies upon raising the temperature of fresh concrete to at least
32 °C (90 °F). The long-term strength is consequently reduced by 10 to 20 per cent
compared with normally cured concrete, but formwork can be removed at the age of
several hours. The rise in temperature is achieved either by heating the aggregate, and
also the water, or by injection of steam into the mixer. In either case, care is required
in controlling the total water content of the mix. Heated or insulated formwork is
necessary.
There are several methods of electrical curing. In one, electric current passes
through the fresh concrete between external electrodes. The current must be
alternating as direct current would lead to hydrolysis of the cement paste. In another
method, a large current at low voltage is passed through the reinforcement in the
concrete member. In a third method, large electric blankets are used to heat the
surface of slabs. Yet another method utilizes insulated resistance wires embedded in
the concrete member; after curing, they are cut and left in the concrete.
Infrared-radiation curing is used in some countries.
Steel formwork can also be heated electrically or through circulating hot water or
oil.
The various specialized curing methods are discussed in ACI 517.2R-878.27 and in
some other publications.8.35,8.36,8.37
Thermal properties of concrete
The thermal properties of concrete are of interest for a variety of reasons, examples of
which are given below. Thermal conductivity and diffusivity are relevant to the
development of temperature gradients, thermal strains, warping, and cracking in the
very early life of concrete, and are also relevant to thermal insulation provided by
concrete in service. Knowledge of thermal expansion of concrete is required in the
design of expansion and contraction joints, in the provision of bridge support
movement, both horizontally and vertically, and in the design of statically
indeterminate structures subject to temperature variation. This knowledge is also
required in the assessment of thermal gradients in concrete, and in the design of
prestressed concrete members. Behaviour at high temperatures needs to be known in
special applications and also in consideration of the effects of fire. Thermal effects in
mass concrete are of especial interest and will be discussed in a later section.
Thermal conductivity
This measures the ability of the material to conduct heat and is defined as the ratio of
the flux of heat to temperature gradient. Thermal conductivity is measured in joules
per second per square metre of area of body when the temperature difference is 1 °C
per metre of thickness of the body (Btu per hour per sq. ft when temperature
difference is 1 °F per ft of thickness).
The conductivity of ordinary concrete depends on its composition and, when the
concrete is saturated, the conductivity ranges generally between about 1.4 and 3.6
J/m2s °C/m (0.8 to 2.1 Btu/ft2h °F/ft).8.10 Density does not appreciably affect the
conductivity of ordinary concrete but, due to the low conductivity of air, the thermal
conductivity of lightweight concrete varies with its density8.87 (see Fig. 13.16). Typical

405
values of conductivity are listed in Table 8.1. More extensive data have been reported
by Scanlon and McDonald,8.10 and also in ACI 207.1R.8.53 From Table 8.1, it can be
seen that the mineralogical character of the aggregate greatly affects the conductivity
of the concrete made with it. In general terms, basalt and trachyte have a low
conductivity, dolomite and limestone are in the middle range, and quartz exhibits the
highest conductivity, which depends also on the direction of heat flow relative to the
orientation of the crystals. In general, crystallinity of rock increases its conductivity.

Table 8.1. Typical Values of Thermal Conductivity of Concrete (selected from

ref. 8.10)

The degree of saturation of concrete is a major factor because the conductivity of

air is lower than that of water. For instance, in the case of lightweight concrete, an
increase in moisture content of 10 per cent increases conductivity by about one-half.
On the other hand, the conductivity of water is less than half that of the hydrated
cement paste, so that the lower the water content of the mix the higher the
conductivity of the hardened concrete.
A frequent practical difficulty is to know the actual moisture content of the
concrete. Loudon and Stacey8.97 assumed to be typical the values of moisture content in
per cent by volume shown at the top of Table 8.2 and, on that basis, recommended the
use of the values of conductivity given in Table 8.2.

Table 8.2. Values of Thermal Conductivity Recommended by Loudon and

Stacey8.97

406
 Conductivity is little affected by temperature in the region of room temperature. At
higher temperatures, the variation in conductivity is complex. It increases slowly with
an increase in temperature up to a maximum at about 50 to 60 °C (122 to 140 °F).
With the loss of water from the concrete as the temperature increases to 120 °C
(248 °F), conductivity decreases sharply. At temperatures in excess of 120 to 140 °C
(248 to 284 °F), the value of conductivity tends to stabilize:8.37 at 800 °C (1470 °F) it is
about one-half of the value at 20 °C (68 °F).8.98
Thermal conductivity is usually calculated from the diffusivity, the latter being
easier to measure, but a direct determination of conductivity is of course possible.
However, the method of test may affect the value obtained. For instance, the steady-
state methods (hot plate and hot box) yield the same thermal conductivity for dry
concrete, but give too low a value for moist concrete because the temperature gradient
causes migration of moisture. For this reason, it is preferable to determine the
conductivity of moist concrete by transient methods; the hot wire test has been found
successful.8.99
Thermal diffusivity
Diffusivity represents the rate at which temperature changes within a mass can take
place, and is thus an index of the facility with which concrete can undergo

407
temperature changes. Diffusivity, δ, is simply related to the conductivity K by the
equation:

where c is the specific heat, and ρ is the density of concrete.

From this expression, it can be seen that conductivity and diffusivity vary in step.
Because of this direct relation, diffusivity is affected by the moisture content of the
concrete, which depends on the original water content of the mix, degree of hydration
of cement, and exposure to drying.
The range of typical values of diffusivity of ordinary concrete is between 0.002
and 0.006 m2/h (0.02 to 0.06 ft2/h), depending on the type of aggregate used. The
following rock types are listed in order of increasing diffusivity: basalt, limestone, and
quartzite.8.10
The measurement of diffusivity consists essentially of determining the relation
between time and the temperature differential between the interior and the surface of a
concrete specimen (both initially at the same temperature) when a change in
temperature is introduced at the surface. Details of procedure and calculation are
given in U.S. Bureau of Reclamation Procedure 4909-92.8.8 Because of the influence
of moisture in the concrete on its thermal properties, diffusivity should be measured
on specimens with a moisture content which will exist in the actual structure.
Specific heat
Specific heat, which represents the heat capacity of concrete, is little affected by the
mineralogical character of the aggregate, but is considerably increased by an increase
in the moisture content of the concrete. Specific heat increases with an increase in
temperature and with a decrease in the density of the concrete.8.110 The common range
of values for ordinary concrete is between 840 and 1170 J/kg per °C (0.20 and 0.28
Btu/lb per °F). The specific heat of concrete is determined by elementary methods of
physics.
Another thermal property of concrete, which is of interest in consideration of fire
effects, is thermal absorptivity. This is defined as (Kρc)1/2, where K is thermal
conductivity, ρ is the density, and c is the specific heat. The thermal absorptivity of
normal weight concrete was reported8.33 as 2190 J/m2 s1/2 per °C (6.44 Btu/ft2 h1/2 per °F).
For lightweight concrete with a density of 1450 kg/m3 (90.5 lb/ft3), the value is 930
J/m2 s1/2 per °C (2.73 Btu/ft2 h1/2 per °F).
Coefficient of thermal expansion
Like most engineering materials, concrete has a positive coefficient of thermal
expansion, but its value depends both on the composition of the mix and on its hygral
state at the time of the temperature change.
The influence of the mix proportions arises from the fact that the two main
constituents of concrete, hydrated cement paste and aggregate, have dissimilar
thermal coefficients, and the coefficient of concrete is a resultant of the two values.
The linear coefficient of thermal expansion of hydrated cement paste varies between
about 11 × l0–6 and 20 × 10–6 per °C (6 × l0–6 and 11 × 10–6 per °F),8.88 and is higher than
the coefficient of aggregate. In general terms, the coefficient of concrete is a function
of the aggregate content in the mix (Table 8.3) and of the coefficient of the aggregate

408
itself.8.89 The influence of the latter factor is apparent from Fig. 8.11, and Table
8.4 gives the values of the coefficient of thermal expansion of 1:6 concretes made
with different aggregates.8.90 The significance of the difference between the
coefficients of the aggregate and the hydrated cement paste was discussed on p. 148.
Here, it can be added that this difference8.5,8.34 may have a deleterious effect when
combined with other actions. Thermal shock which produces a temperature
differential of 50 °C (90 °F) between the surface of concrete and its core has been
reported to cause cracking.8.114

Table 8.3. Influence of Aggregate Content on the Coefficient of Thermal

Expansion8.94

409
Fig. 8.11. Influence of the linear coefficient of thermal expansion of aggregate on
the coefficient of thermal expansion of a 1:6 concrete8.90 (Crown copyright)

Table 8.4. Coefficient of Thermal Expansion of 1:6 Concretes made with

Different Aggregates8.90 (Crown copyright)

410
 The influence of the moisture condition applies to the paste component and is due
to the fact that the thermal coefficient is made up of two parts: the true kinetic
coefficient and swelling pressure. The latter arises from a decrease in the capillary
tension of water held by the hydrated cement paste8.91 and in the adsorbed water in it,
with an increase in temperature.8.40
The moisture-dependent part of the coefficient of thermal expansion does not
include the movement of free water out of, or into, the concrete which results,
respectively, in shrinkage and swelling. Because the moisture-related response to
changes in temperature takes time, the resulting part of the coefficient of thermal
expansion can be determined only when equilibrium has been reached. No swelling is
possible, however, when the cement paste is dry, i.e. the capillaries are unable to
supply water to the gel. Likewise, when the hydrated cement paste is saturated, no
capillary menisci exist and there is therefore no effect of a change in temperature. It
follows that, at these two extremes, the coefficient of thermal expansion is lower than
when the paste is partially saturated. When the paste is self-desiccated, the coefficient
is higher because there is not enough water for a free exchange of moisture to occur
between capillary and gel pores after the temperature change.
When saturated paste is heated, the moisture diffusion from gel to capillary pores,
at a given gel water content, is partially offset by contraction as gel loses water so that
the apparent coefficient is smaller.8.100 Conversely, on cooling, the contraction due to
moisture diffusion from capillary to gel pores, at a given gel water content, is partially
offset by expansion which occurs when the gel absorbs water.8.100
Actual values are shown in Fig. 8.12, and it can be seen that, for young pastes, the
coefficient is a maximum at a relative humidity of about 70 per cent. The relative
humidity at which the coefficient is a maximum decreases with age, down to about 50
per cent for very old hydrated cement pastes8.88 (Fig. 8.13). Likewise, the coefficient
itself decreases with age owing to a reduction in the potential swelling pressure due to
an increase in the amount of ‘crystalline’ material in the hardened paste. Using
saturated concrete, Wittmann and Lukas8.107 confirmed the decrease in the coefficient
with age when the temperature is above freezing. No such variation in the coefficient

411
of thermal expansion is found in high-pressure steam-cured cement paste because it
contains no gel (Fig. 8.12). Only the values determined on saturated or desiccated
specimens can be considered to represent the ‘true’ coefficient of thermal expansion,
but it is the values at intermediate humidities that are applicable to many concretes
under practical conditions.

Fig. 8.12. Relation between ambient relative humidity and the linear coefficient
of thermal expansion of neat cement paste cured normally and high-pressure
steam cured8.88

412
 Fig. 8.13. The linear coefficient of thermal expansion of neat cement paste at
different ages8.88
The chemical composition and fineness of cement affect the thermal expansion
only in so far as they influence the properties of gel at early ages. The presence of air
voids is not a factor.
Figures 8.12 and 8.13 refer to neat cement pastes but the effects are apparent also
in concrete; here, though, the variation in the coefficient is smaller as only the paste
component is affected by the relative humidity and ageing. Measurements of the
coefficient of thermal expansion of concrete in a beam outdoors have confirmed that
the coefficient varies with the moisture content of concrete and is higher (by perhaps
as much as 10–6 per °C) when the concrete is drying.8.39 For the same concrete, the
coefficient of thermal expansion was found to be 11 × 10–6 per °C in winter and 13 ×
10–6 per °C in summer.8.39
Table 8.4 gives values of the coefficient for 1:6 concretes: cured in air at 64 per
cent relative humidity, saturated (water-cured), and wetted after air-curing. A method
of determination of the linear coefficient of thermal expansion of oven-dry ‘chemical-
resistant’ mortar is given in ASTM C 531-00 (2005) and of saturated concrete in U.S.
Corps of Engineers Standard CRD-C 39-81.8.30
The data considered so far apply only at temperatures above freezing and below,
say, 65 °C (150 °F). Considerably higher temperatures can, however, be encountered
in some industrial applications and in airfield pavements used by vertical take-off
aircraft where concrete temperature of 350 °C (660 °F) was recorded.8.38 Before
commenting on the effect of high temperatures on the coefficient of thermal
expansion of concrete, it is useful to note that the coefficient of neat cement paste
decreases above a temperature of about 150 °C (300 °F) and becomes negative above
a temperature of 200 to 500 °C; a value of –32.8 × 10–6 per °C was reported.8.32 The
change in the sign of the coefficient occurs at a lower temperature when the increase
in temperature occurs slowly.8.32 The reason for this is the loss of water from the
hydrated cement paste and possibly internal collapse. The aggregate, however, has a
positive coefficient of thermal expansion at all temperatures and this effect dominates
the expansion of concrete, which expands with an increase in temperature up to high
values. Values of the coefficient of thermal expansion at high temperatures are listed
in Table 8.5.8.92

Table 8.5. Coefficient of Thermal Expansion of Concrete at High

Temperatures8.92

413
 At the other extreme, temperature near freezing results in a minimum positive
value of the coefficient of thermal expansion; at still lower temperatures, the
coefficient is higher again, and indeed somewhat higher than at room
temperature.8.107 Figure 8.14 shows the values of the coefficient for saturated hydrated
cement paste tested in saturated air. In concrete slightly dried after a period of initial
curing and then stored at a relative humidity of 90 per cent and tested at that humidity,
the decrease in the coefficient of thermal expansion at low temperatures is absent (Fig.
8.14).

414
 Fig. 8.14. Relation between the linear coefficient of thermal expansion and
temperature of hydrated cement paste specimens (with a water/cement ratio of
0.40) stored and tested at the age of 55 days under different conditions of
humidity8.107
Laboratory tests have shown that concretes with a higher coefficient of thermal
expansion are less resistant to temperature changes than concretes with a lower
coefficient.8.89 Figure 8.15 shows the results of tests on concrete heated and cooled
repeatedly between 4 and 60 °C (40 and 140 °F) at the rate of 2.2 °C (4 °F) per minute.
However, the data are not sufficient for the coefficient of thermal expansion to be
considered as a quantitative measure of durability of concrete subjected to frequent or
rapid changes in temperature (cf. p. 148).

415
 Fig. 8.15. Relation between the linear coefficient of thermal expansion of
concrete and the number of cycles of heating and cooling required to produce a
75 per cent reduction in the modulus of rupture8.89
Nevertheless, rapid changes in temperature, generally faster than encountered
under normal conditions, may lead to deterioration of concrete: Fig. 8.16 shows the
effects of quenching after heating to the indicated temperature.8.93

416
 Fig. 8.16. Effect of the rate of cooling on the strength of concrete made with a
sandstone aggregate and previously heated to different temperatures8.93
Strength of concrete at high temperatures and resistance to fire
Reports on tests intended to establish the effect of exposure to high temperature, up to
about 600 °C (1100 °F), give widely varying results. The reasons for this include:
differences in the stress acting upon, and in the moisture condition of, the concrete
while being heated; differences in the length of exposure to the high temperature; and
the differences in the properties of the aggregate. In consequence, globally valid
generalizations are difficult. Moreover, the knowledge of the strength of concrete may
be required for different practical conditions of exposure; for instance, in the case of
fire, the exposure to the high temperature is only of a few hours’ duration but the heat
flux is large and so is the mass of concrete subjected to it. Conversely, in cutting
concrete by a thermic lance, the exposure to high temperature is only of a few seconds’
duration and the heat flux applied is very low. In what follows, test data from several
investigations will be referred to, and these have to be interpreted in the light of the
foregoing comments. The adverse effects are due to decomposition of the CSH gel
and of C2S.8.116
The compressive and splitting tensile strengths of concrete, made with limestone
aggregate, exposed to a high temperature for 1 to 8 months are shown in Fig.

417
8.17.8.45 The specimens tested were 100 mm by 200 mm (4 in. by 8 in.) cylinders,
moist-cured for 28 days, then stored in the laboratory for 16 weeks. They were then
heated at the rate of up to 20 °C per hour (36 °F per hour) under conditions such that
loss of water from the concrete could take place. From Fig. 8.17, it can be seen that,
relative to the strength prior to the exposure to the high temperature, there is a steady
loss in strength with an increase in temperature. The relative loss in compressive
strength is very slightly smaller at the water/cement ratio of 0.60 than at the
water/cement ratio of 0.45; this trend does not necessarily continue down to the
water/cement ratio of 0.33.8.42 However, leaner mixes appear to suffer a relatively
lower loss of strength than richer ones.8.95

Fig. 8.17. Influence of exposure to a high temperature on the compressive and

splitting-tensile strengths of concrete, made with a water/cement ratio of 0.45,
expressed as a percentage of strength before exposure (based on ref. 8.45)
The influence of the water/cement ratio on the loss of strength is not noticeable in
the splitting tensile strength; the loss in this strength is similar to that in the
compressive strength.8.45 It can be added that no effect of the length of exposure
(between 1 and 8 months) was observed. Also, there was no difference in the relative
loss of strength between concrete made with Portland cement only and concrete
containing fly ash or ground granulated blastfurnace slag.8.45
Further tests by the same researchers8.42 have shown that an increase in the length of
exposure to a temperature of 150 °C (302 °F) or higher, from 2 to 120 days, increases
the loss of compressive strength. However, the major part of the loss occurs
early.8.42 Tests8.44 on concrete with basalt aggregate showed that the major part of the
loss of strength occurs within 2 hours of the rise in temperature. It should be noted,
however, that the exposure temperature is not necessarily the same as the temperature
within the concrete so that it has to be emphasized once again that the details of the
test method influence the measured output of the tests, but these details cannot always
be fully appreciated from the published description of the tests. All these factors lead

418
to a broad band of the loss of strength as a function of temperature, as shown in Table
8.6.

Table 8.6. Compressive Strength as a Percentage of 28-day Strength at Room

Temperature (based on ref. 8.44)

Lightweight aggregate concrete exhibits a much lower loss of compressive

strength than normal weight concrete: a residual strength of at least 50 per cent after
exposure to 600 °C was reported.8.112
Tests8.48 on high strength concrete (89 MPa) suggest a higher relative loss of
strength than is the case with normal strength concrete. What is more important with
respect to high performance concrete, which contains silica fume, is the occurrence of
explosive spalling associated with high temperature. This was observed by Hertz8.47 in
concrete heated to temperatures in excess of about 300 °C (570 °F) even at a
relatively slow rate of rise in temperature of 60 °C per hour, which is an order of
magnitude lower than in a fire. Explosive spalling was confirmed in tests on concrete
containing silica fume and having a water/cement ratio of 0.26.8.43 This might seem
surprising as the volume of water involved is small but, on the other hand, the
permeability is extremely low.
It can be stated more generally that the risk of explosive spalling is higher the
lower the permeability of the concrete and the higher the rate of rise in temperature.
An associated observation is that the loss in strength at higher temperatures is greater
in saturated than in dry concrete, and it is the moisture content at the time of
application of load that is responsible for the difference.8.101
The influence of moisture content on strength is apparent also in fire tests on
concrete, where excessive moisture at the time of fire is the primary cause of spalling.
In general, moisture content of the concrete is the most important factor determining
its structural behaviour at higher temperatures.8.111 In massive concrete members,
moisture movement is extremely slow so that the effects of a high temperature, while
loss of water is prevented, may be more serious than in thin members. Inclusion of
polypropylene fibres in the mix is helpful.
One of the changes which occurs as the temperature rises to about 400 °C (or
750 °F) is the decomposition of calcium hydroxide so that lime is left behind in
consequence of drying.8.7 If, however, after cooling, water ingresses into concrete, the
re-hydration of lime can be disruptive; thus the damage manifests itself subsequently
to the fire. From this standpoint, inclusion of pozzolanas in the mix, which remove
calcium hydroxide, is beneficial.
While it is the behaviour of concrete that is of practical interest, the overall
behaviour of concrete may mask some of the changes which occur in small specimens
of hydrated cement paste. Tests8.46 on paste specimens having a water/cement ratio of
0.30 and wet-cured for 14 weeks, heated and tested in compression while hot, showed
a decrease in strength with an increase in temperature up to 120 °C (248 °F). At

419
higher temperatures, the strength was found to be approximately equal to the original
value. This strength is maintained up to 300 °C (572 °F). However, at still higher
temperatures, there is a severe and progressive decrease in strength. The unimpaired
strength at intermediate temperatures is ascribed by Dias et al.8.46 to the disappearance
of the disjoining pressure (see p. 37) and densification of the gel. In concrete, such
changes would be limited by the difficulty of effective drying.
Modulus of elasticity at high temperatures
The behaviour of structures is often dependent on the modulus of elasticity of the
concrete, and this modulus is strongly affected by temperature. The pattern of
influence of temperature on the modulus of elasticity is shown in Fig. 8.18. For mass-
cured concrete, there is no difference in modulus in the range of 21 to 96 °C (70 to
205 °F),8.102 but the modulus of elasticity is reduced at temperatures in excess of
121 °C (250 °F).8.56 However, when water can be expelled from concrete, there is a
progressive decrease in the modulus of elasticity between about 50 and 800 °C (120
and 1470 °F) (see Fig. 8.18);8.43,8.104 relaxation of bonds may be a factor in this. The
extent of the decrease in the modulus depends on the aggregate used, but a
generalization on this subject is difficult. In broad terms, the variation of strength and
of modulus with temperature is of the same form.

Fig. 8.18. Influence of temperature on modulus of elasticity of concrete (based on

refs 8.48 and 8.104)
Behaviour of concrete in fire
Although reference to fire was made on several occasions, the full treatment of the
resistance of concrete to fire is a topic outside the scope of this book because fire
endurance applies really to a building element rather than to a building material. We
can say, however, that, in general, concrete has good properties with respect to fire
resistance; that is, concrete is non-combustible, the period of time under fire during
which concrete continues to perform satisfactorily is relatively high, and no toxic
fumes are emitted. The relevant criteria of performance are: load-carrying capacity,
resistance to flame penetration, and resistance to heat transfer when concrete is used
as a protective material for steel. A general review of the resistance of concrete to fire
has been written by Smith.8.6

420
 In practice, what is required of structural concrete is that it preserves structural
action over a desired length of time (known as fire rating). This is distinct from being
heat resisting.8.78 Considering the behaviour of concrete as a material, we should note
that fire introduces high temperature gradients and, as a result, the hot surface layers
tend to separate and spall from the cooler interior of the body. The formation of
cracks is encouraged at joints, in poorly compacted parts of the concrete, or in the
planes of reinforcing bars; once the reinforcement has become exposed, it conducts
heat and accelerates the action of high temperature.
The type of aggregate influences the response of concrete to high temperature. The
loss of strength is considerably lower when the aggregate does not contain silica
(some forms of which undergo change), e.g. with limestone, basic igneous rocks, and
particularly with crushed brick and blastfurnace slag. Concrete with a low thermal
conductivity has a better fire resistance so that, for instance, lightweight concrete
stands up better to fire than ordinary concrete.
It is interesting to note that dolomitic gravel leads to a very good fire resistance of
concrete. The reason for this is that the calcination of the carbonate aggregate is
endothermic;8.103 in consequence, heat is absorbed and a further temperature rise is
delayed. Also, the calcined material has a lower density and therefore provides a
measure of surface insulation. This effect is significant in thick members. On the
other hand, if pyrites is present in the aggregate, slow oxidation at about 150 °C
(330 °F) causes disintegration of the aggregate and consequently rupture of the
concrete.8.42
Abrams8.108 confirmed that, at temperatures above about 430 °C (810 °F), siliceous
aggregate concrete loses a greater proportion of its strength than concretes made with
limestone or lightweight aggregates but, once the temperature has reached some
800 °C (1470 °F), the difference disappears (Fig. 8.19). For practical purposes, about
600 °C (1100 °F) can be considered as the limiting temperature for structural integrity
of concrete made with Portland cement; at higher temperatures, refractory concrete
has to be used (see p. 102). The relevant temperature is that of the concrete itself and
not the temperature of the flame or of gases. Sullivan8.117 showed that explosive
spalling occurs especially at a w/c of 0.35 because the concrete has a particularly low
surface permeability (see also p. 689.)

421
 Fig. 8.19. Reduction in compressive strength of concrete heated without
application of load and then tested hot; average initial strength of 28 MPa (4000
psi)8.108
With all aggregates, the percentage loss of strength was found to be independent of
the original level of strength but the sequence of heating and loading influences the
residual strength. Specifically, concrete heated under load retains the highest
proportion of its strength, whereas heating unloaded specimens leads to the lowest
strength of the subsequently cooled concrete. Application of load while the concrete is
still hot leads to intermediate values. Typical results are shown in Fig. 8.20 (Figure
2.9 may be also of interest.)

422
 Fig. 8.20. Reduction in compressive strength of concrete made with limestone
aggregate: (A) heated without application of load and then tested hot; (B) heated
under an initial stress/strength ratio of 0.4 and then tested hot; (C) heated
without application of load and tested after 7 days of storage at 21 °C (70 °F)8.108
Application of water in a fire is tantamount to quenching: this causes a large
reduction in strength because severe temperature gradients are set up in the concrete.
Concretes made with siliceous or limestone aggregate show a change in colour
with temperature. As this change is dependent on the presence of certain compounds
of iron, there is some difference in the response of different concretes. The change in
colour is permanent, so that the maximum temperature during a fire can be
estimated a posteriori. The colour sequence is approximately as follows: pink or red
between 300 and 600 °C, then grey up to about 900 °C, and buff above
900 °C.8.93 Thus, the residual strength can be approximately judged: generally,

423
concrete whose colour has changed beyond pink is suspect, and concrete past the grey
stage is probably friable and porous.8.1
Attempts have been made to determine the maximum temperature which concrete
had reached during a fire by measuring the reduction in thermoluminescence. This is a
light signal which is a function of temperature. However, the light output is affected
by the length of exposure to the high temperature so that the reduction in the strength
of concrete exposed to fire for a prolonged period can be significantly
underestimated.8.41
Deliberate application of very high temperature over a small area is used in flame
cleaning of concrete surfaces. This does not damage the concrete beyond the depth
removed, which is 1 to 2 mm (0.04 to 0.08 in.) provided the blowpipe is moved at the
requisite rate.8.109 Under such circumstances, even though the flame temperature is
about 3100 °C, the maximum concrete temperature is not more than 200 °C.
Strength of concrete at very low temperatures
The development of strength of concrete at temperatures higher than –11 °C (12 °F)
was considered on p. 305, this being the lowest temperature at which hydration takes
place and a gain in strength occurs. There exist, however, practical situations of
exposure to cryogenic temperatures of concrete which has hardened at room
temperature; this is, for instance, the case in storage tanks for liquefied natural gas
whose boiling point is –162 °C (–260 °F). The effect of these very low temperatures
will now be considered.
At temperatures ranging from the freezing point of water down to about –200 °C
(–330 °F), the strength of concrete is markedly higher than at room temperature. The
compressive strength may be as high as two to three times the strength at room
temperature when the concrete is moist while being chilled, but the compressive
strength of air-dry concrete increases very much less.
The difference in the increase in strength between wet and dry concretes is related
to the formation of ice in the hydrated cement paste. The freezing point of gel water is
lower the smaller the pore size so that all the adsorbed water becomes frozen at a
temperature between –80 and –95 °C (–112 and –139 °F). As ice can resist stress,
unlike the water which it replaces, frozen concrete has an extremely low effective
porosity and, therefore, high strength. The strength of ice and its coefficient of
thermal expansion vary with temperature so that the changes occurring in hydrated
cement paste are complex.849
If the concrete is not exposed to low temperature, empty pores remain empty so
that the increase in strength is small.
The pattern of the relation between compressive strength and temperature, both for
moist and for air-dry lightweight aggregate concrete, is shown in Fig. 8.21. The
corresponding data for splitting-tensile strength are shown in Fig. 8.22. From this
figure, it can be seen that the increase in the tensile strength occurs mainly between –7
and –87 °C. Also the relative increase in the tensile strength of air-dry concrete is
smaller than the relative increase in compressive strength. The data in Figs
8.21 and 8.22 refer to lightweight aggregate concrete which, for cryogenic purposes,
has the advantage of good insulating properties. However, in normal weight concrete,
the increase in strength at low temperatures is greater than is the case with lightweight
aggregate concrete.

424
 Fig. 8.21. Effect of very low temperatures on compressive strength of concrete
(measured on standard cylinders) (based on ref. 8.49)

Fig. 8.22. Effect of very low temperatures on splitting-tensile strength of concrete

(based on ref. 8.49)

425
 The pattern of the increase in compressive strength with an increase in moisture
content is independent of the water/cement ratio; an example of this relation for
concrete at –160 °C is shown in Fig. 8.23.8.50 Similar behaviour applies to concrete
with a strength of 80 MPa (12 000 psi) at normal temperature.8.51

Fig. 8.23. Relation between increase in compressive strength at –160 °C (–256 °F)
above the strength at room temperature and moisture content for concretes with
water/cement ratios of 0.45 and 0.55 (based on ref. 8.51)
Figure 8.21 shows that there is little, if any, further increase in compressive
strength when the temperature drops below about –120 °C. The reason for this is that,
in the region of that temperature, changes occur in the structure of ice. Specifically, at
–113 °C, ice changes from hexagonal to orthorhombic structure; this change is
accompanied by a decrease in volume of about 20 per cent. The pattern of strain
development with a decrease in temperature and the behaviour of concrete under
cyclic temperature have been extensively studied by Miura.8.50 It should be noted that
the effects of temperature gradients and of temperature cycling need to be considered
in structural design.

426
 The modulus of elasticity of moist concrete increases steadily with the decrease in
temperature down to –190 °C. At that temperature, the modulus of elasticity is about
1.75 times the modulus at room temperature; for air-dry concrete the corresponding
value is about 1.65.8.49
Mass concrete
In the past, the term ‘mass concrete’ was applied only to concrete of massive
dimensions, such as gravity dams, but nowadays the technological aspects of mass
concrete are relevant to any concrete member of such dimensions that the thermal
behaviour may lead to cracking unless appropriate measures are taken. The crucial
feature of mass concrete is, thus, its thermal behaviour, a design objective for such
concrete being to avoid or to reduce and control the width and spacing of cracks.
It can be recalled from Chapter 1 that hydration of cement generates heat which
causes a rise in the temperature of the concrete. If this rise occurred uniformly
throughout a given concrete element without any external restraint, then the element
would expand until the maximum temperature has been reached; thereafter, as the
concrete cools due to the loss of heat to the ambient atmosphere, uniform contraction
would occur. Thus, there would be no thermal stresses within the element. In practice,
however, restraint exists in all but the smallest of concrete members. There are two
categories of restraint: internal and external.
Internal restraint arises from the fact that, when the surface of concrete can lose
heat to the atmosphere, there develops a temperature differential between the cool
exterior and the hot core of the concrete element, the heat not being dissipated to the
outside fast enough in consequence of the low thermal diffusivity of the concrete. As
a result, the free thermal expansion is unequal in the various parts of the concrete
element. Restraint of the free expansion results in stresses, compressive in one part of
the element and tensile in the other. If the tensile stress at the surface of the element
due to the expansion of the core exceeds the tensile strength of concrete, or if it results
in the tensile strain capacity being exceeded (see p. 294), then surface cracking will
develop.
The actual situation is complex because creep, which is high in very young
concrete, relieves some of the compressive stress induced in the core so that the rate
of change of the temperature is also a factor; this behaviour is discussed on p. 474.
Internal restraint can occur also when concrete is placed against a surface at a
much lower temperature, such as cold ground or uninsulated formwork in cold
weather. In such a situation, different parts of the concrete element set at different
temperatures. When, subsequently, the core of the concrete element cools, its thermal
contraction is restrained by the already cool external part and cracking in the interior
may occur.
Examples of temperature changes are shown in Figs 8.24 and 8.25, which imply
that cracking will occur when the temperature difference exceeds 20 °C (36 °F). This
limit on the temperature difference was suggested by FitzGibbon.8.65,8.66 For a
temperature difference of 20 °C, taking the coefficient of thermal expansion
of concrete as 10 × 10–6 per °C (5.5 × 10–6 per °F) (see Table 8.4), the differential
strain is 200 × 10–6. This is a realistic estimate of tensile strain at cracking (see p. 292).
The following practical experience can be quoted.

427
 Fig. 8.24. An example of the pattern of temperature change which causes
external cracking of a large concrete mass. The critical 20 °C temperature
difference occurs during cooling8.66

428
 Fig. 8.25. An example of the pattern of temperature change which causes
internal cracking of a large concrete mass. The critical 20 °C temperature
difference occurs during heating, but the cracks open only when the interior has
cooled through a greater temperature range than the exterior8.66
In a 1.1 m (43 in.) square column made of reinforced concrete with a Type I
cement content of 500 kg/m3 (840 lb/yd3) and a silica fume content of 30 kg/m3 (50
lb/yd3), a rise in temperature of 45 °C (81 °F) above the ambient temperature was
observed 30 hours after placing.8.52
A similar rise in temperature can occur even in sections with a least dimension of
0.5 m (20 in.). The need for not allowing the surface of the concrete to cool too
rapidly is obvious so that the insulating properties of the formwork and the time of its
removal have to be controlled.
The preceding discussion has shown that the major cause of the temperature
differential in a concrete element is the generation of heat by the hydration of cement.
This topic was discussed on p. 38 in so far as the heat of hydration of a unit mass of
different cements is concerned. It is, therefore, possible to choose a Portland cement
with a chemical composition which leads to a low rate of heat development. However,
with blended cements, an estimate of the heat hydration is more complicated.
Moreover, from the standpoint of the development of a temperature differential, it is
not only the total heat of hydration but also its rate of development that are relevant. It

429
should be remembered that a higher fineness of cement leads to more rapid hydration,
so that it might be desirable to avoid cements with a high specific surface.
The choice of cement, however, offers only a partial solution because it is the
cement content per cubic metre of concrete that largely governs the heat generated.
The remedy, therefore, lies in using a low cement content as well as in using blended
cements because it is the Portland cement that is responsible for early heat generation,
pozzolanas reacting chemically more slowly. It follows that, using a low content of
blended cement with a high proportion of pozzolanas, the maximum temperature rise
can be reduced and its occurrence can be delayed. The benefit of the delay is that the
concrete will have a higher tensile strength and be less prone to cracking.
With any cement, the rate of hydration is higher at higher temperatures so that
cooling the fresh concrete below the ambient temperature (see next section) and
placing it at a low temperature reduces the rate of generation of heat; in addition, the
difference between the maximum temperature of the concrete and the final ambient
temperature is reduced.
In large plain concrete structures, the use of aggregate with a large maximum size,
75 mm (3 in.) or even 150 mm (6 in.), may be desirable because this allows a
reduction in the water content of the mix for a given workability. At a fixed
water/cement ratio, the cement content can, therefore, be reduced. The water/cement
ratio can be high (up to 0.75) because, in structures such as gravity dams, the strength
of concrete is of little structural importance, prevention of cracking and durability
being critical. In any case, it is the strength at greater ages that is likely to be of
relevance. Mixes with a blended cement content of 109 kg/m3 (184 lb/yd3), of which
67 per cent was pozzolana, have been used; the water content was 48 kg/m3 (80 lb/yd3),
the slump 40 mm ( in.), and the 28-day cylinder was strength 14 MPa (2000
psi).8.67 We can note that using a very low cement content is not only economical per
se but leads also to economy in other measures used to overcome the undesirable
effects of the heat of hydration of the cement, such as cooling the concrete in situ by
circulating chilled water through embedded pipework.8.67
We can add that some recent dams have been built using roller-compacted
concrete with a cement content as low as 66 kg/m3 (112 lb/yd3) of which 30 per cent
was fly ash.8.54 However, this specialized material and the associated technology are
outside the scope of the present book.
Let us now consider reinforced concrete: here a much higher strength is required,
often at 28 days, and the use of large-size aggregate may be impractical because of
reinforcement spacing or because obtaining such aggregate may be uneconomical.
Also, embedding pipework may not be permitted. The essential problem is,
nevertheless, the same as in plain concrete, i.e. the interior of the mass will heat up
more than the exterior if the loss of heat at the surface is large. If the difference in
temperature between the interior and the exterior is large enough, cracking will
develop. However, appropriate detailing of the reinforcement can control the width
and spacing of the cracks. FitzGibbon8.65,8.66 estimated that the temperature rise under
adiabatic conditions is 12 °C per 100 kg of cement per cubic metre of concrete (13 °F
per 100 lb/yd3), regardless of the type of cement used, for cement contents between
300 and 600 kg/m3 (500 and 1000 lb/yd3).
The solution to the problem is not to limit the temperature rise in the interior but
rather to prevent the heat loss at the surface. Thus, the entire concrete mass is allowed
to heat, more or less to the same degree, and expand without restraint; with time,

430
cooling, again more or less uniform throughout, takes place, and the structure reaches
its final dimensions, again without restraint. To prevent a large heat loss, the
formwork and the top surface of the structure must be adequately insulated with
polystyrene or urethane; additional insulation is needed at edges and corners where
the heat loss occurs in more than one direction and in other sensitive parts of the
structure.
In practice, the temperature at various points should be monitored by thermo-
couples, and insulation should be adjusted accordingly. The insulation must control
loss of heat by evaporation, conduction, and radiation. To achieve the first, a plastic
membrane or a curing compound should be used, but not spraying or ponding as these
have a cooling effect. Plastic-coated quilts are useful in all respects but softboard can
also be used. The insulation must be maintained until the temperature differential has
been reduced to 10 °C (18 °F).
Other specialized measures are also required in order to achieve a monolithic
structure without cold joints. One measure is a differential use of retarders so that the
concrete in the lower part remains plastic until completion of placing, possibly in 12
hours; bleeding also needs to be controlled. One of the largest continuous pours to
date is that of a reinforced concrete foundation containing 12 000 m3 (16 000 yd3) of
concrete.8.53
It is useful to point out that care is required if concretes with dissimilar thermal
properties are placed so as to create a monolithic element. An example of this is a
highway slab placed in two layers (so that steel dowels in the contraction joint can be
inserted) which contain different blended cements.8.2
External restraint of thermal movement can result in cracking of reinforced
concrete members, even when thin. This is the case with walls cast onto an existing
foundation which restrains the thermal movement due to the rise in the temperature of
the wall concrete: vertical cracks through the full thickness of the wall at its base can
extend a considerable distance upwards. Prevention of cracking can be achieved by
appropriate structural detailing of the reinforcement, but an understanding of the
thermal behaviour of concrete is essential to reduce the severity of the problem.
The preceding extensive discussion of the temperature rise in a concrete mass has
shown that the temperature depends on the position in the concrete element, as well as
on the age of the concrete, and on the details of the insulation. The properties of
concrete at a particular position can be determined by the use of temperature-matched
curing. This is a technique in which a thermocouple inserted into concrete at a
specified position controls the temperature of a bath in which a concrete specimen is
placed; the specimen is isolated from the water. The properties of the temperature-
matched concrete which are of greatest interest are strength and creep. The knowledge
of strength can be used to determine formwork striking times or the transfer of
prestress. Creep is of relevance in structural design.
The determination of temperature in different locations within a concrete mass can
be used to adjust thermal insulation so as to minimize temperature gradients within
the mass.
A particular type of continuous pour of concrete is used in certain types of
structures such as high-rise buildings and towers. Here, the formwork rises
continuously, or nearly so, over a period of days or weeks. This is known as
slipforming.8.115 Horizontal slipforming can be used in the construction of curbs.

431
Concreting in hot weather
There are some special problems involved in concreting in hot weather, arising both
from a higher temperature of the concrete and, in many cases, from an increased rate
of evaporation from the fresh mix. These problems concern the mixing, placing and
curing of the concrete.
Hot-weather concreting is not so much an unusual or a specialized process; rather,
it requires taking certain recognized measures to minimize or control the effects of
high ambient temperature, high temperature of the concrete, low relative humidity,
high wind velocity, and high solar radiation. What is required on each construction
project where any one or more of the above conditions exist is to develop appropriate
techniques and procedures and to follow them rigorously; uniformity is vital and
departures from the established norm spell trouble.
A higher temperature speeds up the setting time of concrete, as defined in ASTM
C 403-08. Tests on a 1:2 cement–sand mortar8.3 showed that the initial setting time was
approximately halved by a change in the temperature of the concrete from 28 to 46 °C
(82 to 115 °F). The effect was similar at water/cement ratios between 0.4 and 0.6, but
the actual setting time was shorter the lower the water/cement ratio.8.3
A high ambient temperature causes a higher water demand of the concrete and
increases the temperature of the fresh concrete. This results in an increased rate of
loss of slump and in a more rapid hydration, which leads to accelerated setting and to
a lower long-term strength of concrete (see p. 359). Furthermore, rapid evaporation
may cause plastic shrinkage cracking and crazing, and subsequent cooling of the
hardened concrete can introduce tensile stresses. It is generally believed that plastic
shrinkage cracking is likely to occur when the rate of evaporation exceeds the rate at
which the bleeding water rises to the surface, but it has been observed that cracks also
form under a layer of water and merely become apparent on drying.8.61 Evaporation
rate in excess of 1.0 kg/m2 per hour (0.2 lb/ft2 per hour) is considered to be critical.8.14
Plastic shrinkage cracks can be very deep, ranging in width between 0.1 and 3 mm
(0.004 and 0.12 in.), and can be quite short or as long as 1 m (or 3 ft).8.62 Once
developed, they are difficult to close permanently.8.14 A drop in the ambient relative
humidity encourages this type of cracking8.9 so that, in fact, the causes of it appear to
be rather complex. According to ACI 305R-918.14 the risk of plastic cracking is the
same at the following combinations of temperature and relative humidity:
41 °C (105 °F) and 90 per cent
35 °C (95 °F) and 70 per cent
24 °C (75 °F) and 30 per cent.
Wind velocity in excess of 4.5 m/s (10 mph) aggravates the situation;8.14 wind shields
are helpful and so is the provision of sun shades.8.20
Another type of cracking on the surface of fresh concrete is caused by differential
settlement of fresh concrete due to some obstruction to settlement, such as large
particles of aggregate or reinforcing bars. This plastic settlement cracking can be
avoided by the use of a dry mix, good compaction, and by not allowing too fast a rate
of build-up of concrete. Plastic settlement cracking can occur also at normal
temperatures but, in hot weather, plastic shrinkage cracking and plastic settlement
cracking are sometimes confused with one another.
There are some further complications in hot-weather concreting: air-entraining is
more difficult, although this can be remedied by using larger quantities of the

432
entraining agent. A related problem is that, if relatively cool concrete is allowed to
expand when placed at a higher temperature, the air voids expand and the strength is
reduced. This would occur, for instance, with horizontal panels but not with vertical
ones in steel moulds where expansion is prevented.8.64
Let us now consider the steps which can be taken to avoid or reduce the ill-effects
of hot weather. In the past, a maximum air temperature at which concrete could be
placed used to be limited. This is not a sensible restriction in countries with very high
ambient temperatures. Nevertheless, a limit on the placing temperature of concrete
which will be exposed to a humid or aggressive environment is 30 °C (86 °F). The
current British standard dealing with hot weather concreting is BS 8500-1 : 2006.
Whenever possible, it is desirable to place the concrete in the coolest part of the day
and preferably at a time such that the ambient temperature will rise following the
setting of the concrete, that is, after midnight or in the early hours of the morning. It is
worth adding that trial batches of the concrete should be made at what is intended to
be the placing temperature, and not at some other temperature such as the laboratory
temperature of 20 or 25 °C (68 or 77 °F).
There are a number of preventive measures that can be taken. In the first instance,
the cement content should be kept as low as possible so that the heat hydration does
not unduly aggravate the effects of high ambient temperature. The temperature of the
fresh concrete can be lowered by pre-cooling one or more of the ingredients of the
mix. A placing temperature of concrete as low as 10 °C (50 °F) is desirable but may
well be impractical.
The temperature T of the freshly mixed concrete can be easily calculated from that
of the ingredients, using the expression

where T denotes temperature in °C or °F, W the mass of ingredient per unit volume of
concrete, and the suffixes a, c, w refer to aggregate, cement, and water (both added
and in aggregate) respectively. The figure of 0.22 is the approximate ratio of the
specific heat of the dry ingredients to that of water, and is applicable to both SI and
Imperial (American) systems of units. It may be worth pointing out that, during the
night, aggregate and water do not cool as rapidly as the air so that their temperature
cannot be assumed to be equal to the air temperature.
The actual temperature of the concrete will be somewhat higher than indicated by
the above expression due to the mechanical work done in mixing, and will further rise
due to the development of the heat of wetting and hydration of cement, as well as due
to the heat transfer from the ambient air and formwork. Incidentally, it is important
that the formwork should be cooled prior to the placing of concrete. To obtain a better
picture, we can say that if the water/cement ratio of a mix is 0.5 and the
aggregate/cement ratio is 5.6, then a drop of 1 °C (or 1 °F) in the temperature of fresh
concrete can be obtained by lowering the temperature either of the cement by 9 °C
(9 °F) or of the water by 3.6 °C (3.6 °F) or of the aggregate by 1.6 °C (1.6 °F). It can
be seen that because of its relatively small quantity in the mix the temperature of the
cement is not important.
The use of hot cement per se is not detrimental to strength but it is preferable not
to use cement at temperatures above about 75 °C (170 °F). This statement is of
interest because hot cement is sometimes viewed with suspicion and various ill effects

433
have at times been ascribed to its use. However, if hot cement is dampened by a small
amount of water before it is well dispersed with other solids it may set quickly and
form cement balls.
There exist various means of cooling the aggregate and the mix water. Coarse
aggregate can be cooled by spraying with chilled water or by inundation. Another
method is to use evaporative cooling by blowing air, preferably chilled, through moist
aggregate. Fine aggregate can also be cooled by air; freezing by liquid nitrogen has
been tried8.19 but the fine aggregate must be surface dry. Precooling of aggregate in a
closed mixer by means of liquefied carbonic acid gas (dry ice), which melts at –78 °C,
has also been tried.8.15
Mix water can be chilled or can be replaced, usually only partially, by crushed or
flaked ice; ice is a highly efficient means of cooling because 1 kg of ice absorbs 334
kJ when melting at 0 °C, which is a quantity of heat four times greater than cooling
the water by 20 °C. All the ice must melt prior to the end of the mixing operation.
Liquid nitrogen, which absorbs 240 kJ/kg when vaporizing at –196 °C, can also be
used to chill the water down to 1 °C or can be injected direct into a stationary mixer
or a truck mixer immediately prior to discharge. The cost of liquid nitrogen, including
the necessary equipment, is high. On the basis of cost per 1 degree drop in
temperature of concrete, the use of heat pumps to cool the water is very
economical8.13 but, of course, applicable only at a static mixing plant. A range of
cooling techniques is described in ACI 207.4R-93,8.4 and ACI 305R-918.14 contains
advice on insulating and painting white the equipment involved in storing the mix
ingredients, and also on mixing and transporting concrete.
After placing, concrete should be protected from the sun; otherwise, if a cold night
follows, cracking is likely to occur, the extent of cracking being directly related to the
temperature difference. In dry weather, wetting concrete and allowing evaporation to
take place results in effective cooling; there is no cooling by this means when
membrane curing is used so that a higher temperature may be reached. Large exposed
areas such as roads and airfields are particularly vulnerable.
Proper curing in hot weather may be of shorter duration because an advanced
degree of hydration is reached more rapidly than at lower temperatures. The emphasis
is on the word ‘proper’ because, as already mentioned, a higher temperature also
promotes more rapid drying of the concrete.8.60
The greatest interest in hot-weather concreting is with respect to hot and dry
conditions. Generalized information about the behaviour and properties of concrete
placed in a hot and continuously humid climate is not available. Data obtained in
specific investigations8.22 show large variations. All we can say is that the absence of
drying in the very early life of concrete is tantamount to the provision of moist curing,
which is beneficial from the standpoint of gradual gain in strength and reduced drying
shrinkage. Nevertheless, the initial high temperature has an adverse effect on long-
term strength. It is also prudent to assume that plastic shrinkage can occur, depending
on the bleeding characteristics of the concrete and on exposure to wind.
Other investigations8.21,8.59 also indicate that the effects of early high temperature are
less detrimental to long-term strength than the absence of moist curing. Great care is
required in translating this observation into practice: whereas wet curing is of
paramount importance, the harmful consequences of early high temperature are also a
reality.
Concreting in cold weather

434
Before discussing the actual concreting operations, we should consider the action of
frost on fresh concrete; the durability of hardened concrete subjected to repeated
cycles of freezing and thawing is discussed in Chapter 11.
In Chapter 6, it was stated that hydration of cement occurs even at low
temperatures down to about –10 °C (14 °F); it is therefore rational to ask: what is then
the significance of the temperature at which water freezes? If concrete which has not
yet set is allowed to freeze, the action of frost is somewhat similar to that in a
saturated soil subject to heaving: the mixing water freezes, with a consequent increase
in the overall volume of the concrete. Furthermore, because no water is available for
chemical reactions, the setting and hardening of concrete are delayed. It follows from
the latter observation that, if concrete freezes immediately after it has been placed,
setting will not have taken place, and thus there is no cement paste that can be
disrupted by the formation of ice. While the low temperature continues, the process of
setting will remain suspended. When, at a later date, thawing takes place, the concrete
should be revibrated, and it will then set and harden without loss of strength. However,
because of the expansion of the mixing water on freezing, a lack of revibration would
allow the concrete to set with a large volume of pores present, and consequently the
strength of the concrete would be very low. Revibration on thawing would produce a
satisfactory concrete, but such a procedure is not recommended except when
unavoidable.
If freezing takes place after the concrete has set but before it has developed an
appreciable strength, the expansion associated with the formation of ice causes
disruption and an irreparable loss of strength. If, however, the concrete has acquired a
sufficient strength, it can resist the freezing temperature without damage, not only by
virtue of the higher resistance to the pressure of the ice but also because a large part of
the mixing water will have become combined with the cement or located in small
pores, and would thus not be able to freeze. It is difficult, however, to establish when
this situation has been reached, because setting and hardening of cement depend on
the temperature during the period preceding the actual advent of freezing. According
to ACI 306R-88,8.55 when concrete has reached a compressive strength of about 3.5
MPa (500 psi), the degree of saturation has dropped below the critical value, provided
no external water has ingressed into the concrete. At that stage, the concrete is capable
of withstanding one cycle of freezing and thawing. Higher values of strength are
recommended in some other countries, but no reliable data are available on the
strength at which concrete can successfully resist temperatures below 0 °C (32 °F).
Generally, the more advanced the hydration of cement and the higher the strength
of concrete the less vulnerable it is to frost. This situation can be expressed by means
of the minimum age of concrete stored at a given temperature when exposure to frost
will not cause damage; typical values (averaged from various sources8.105,8.106) are given
in Table 8.7. Figure 8.26 shows the influence of the age at which first freezing occurs
on the expansion of concrete: the considerable decrease in the magnitude of expansion
of concrete allowed to harden for about 24 hours is noticeable, and protecting
concrete from frost during that period is clearly highly advisable.

Table 8.7. Age of Concrete at Which Exposure to Frost does not Cause Damage

435
Fig. 8.26. Increase in volume of concrete during prolonged freezing as a function
of age when freezing starts8.68
The resistance to alternating freezing and thawing also depends on the age of the
concrete when the first cycle is applied, but this type of exposure is more severe than
prolonged freezing without periods of thaw, and several cycles can cause damage
even to concrete cured at 20 °C (68 °F) for 24 hours.8.68 It may be noted that there is no
direct relation between the frost resistance of young concrete and the durability of
mature concrete subjected to numerous cycles of freezing and thawing,8.69 a topic
which is considered in Chapter 11. Figure 11.2 in that chapter shows a lack of
expansion on the first occasion of freezing when this occurs at the age of more than 1
day: this supports the view expressed in ACI 306R-88 to the effect that most “well-

436
proportioned” concretes stored at 10 °C (50 °F) reach the strength of 3.5 MPa (500 psi)
during the second day.8.55
Concreting operations
When the air temperature is continuously below 0 °C (32 °F), the weather can be
unarguably described as cold. The situation is less clear-cut when there is a large
diurnal variation in temperature. For convenience, the definition of ‘cold weather’
used by ACI 306R-888.55 can be used. This can be paraphrased by saying that cold
weather obtains when two conditions exist: when the average of the maximum and
minimum air temperatures recorded on 3 consecutive days is less than 5 °C (40 °F) as
well as when the air temperature during at least 12 hours in any 24-hour period is
10 °C (50 °F) or lower.
Under such circumstances, normal weight concrete should not be placed
unless its temperature is at least 13 °C (55 °F) for thin sections (300 mm (12 in.)) or at
least 5 °C (40 °F) when the minimum dimension of the concrete element is at least 1.8
m (72 in.).8.55 Lightweight aggregate concrete, which has a lower thermal conductivity,
can be somewhat cooler when placed. Such concrete also has a lower specific heat so
that a given heat of hydration of cement more effectively keeps the lightweight
aggregate concrete from freezing than is the case with normal weight aggregate
concrete.
Advantage can also be taken of using rapid-hardening cement and rich mixes with
a low water/cement ratio, and of the use of cement with a high rate of heat
development, i.e. having high C3S and C3A contents. Accelerators can be used but
chlorides must be avoided if steel is present in the concrete.
To achieve the minimum temperatures cited earlier, when aggregate, water and air
are cold, the mix ingredients can be heated. Water can be heated easily, but it is
inadvisable to exceed a temperature of 60 to 80 °C (140 to 176 °F) because flash set
of the cement may result; the likelihood of this happening depends on the difference
between the temperatures of water and cement. It is important to prevent the cement
from coming into contact with the hot water, and for this reason the order of feeding
the mix ingredients into the mixer must be suitably arranged.
If heating the water does not sufficiently raise the temperature of the concrete, the
aggregate may also be heated. This is done preferably by passing steam through coils
rather than by the use of live steam because the latter methods leads to a variable
moisture content of the aggregate. Heating the aggregate above 52 °C (125 °F) is
inadvisable.8.63 At the other extreme, it is important that the aggregate does not contain
ice as the heat required to melt it would greatly reduce the temperature of the concrete.
The temperature of the mix ingredients must be controlled, and the temperature of
the resulting concrete should be calculated in advance (see p. 401). This calculation
should allow for the loss of heat during the transport of the concrete. The objective is
to ensure that the temperature of the concrete is high enough to prevent premature
freezing but also to make sure that setting does not occur at too high a temperature.
This would adversely affect the development of strength of the concrete (see p. 361).
In addition, a high temperature of fresh concrete lowers its workability and may lead
to high thermal contraction.
It is thus desirable for the concrete to set at, say, 7 to 21 °C (45 to 70 °F). The
temperature of 7 °C (45 °F) applies when the air temperature is no lower than –1 °C
(30 °F) and the concrete element is thick; the value of 21 °C (70 °F) applies when the

437
air is cooler than –18 °C (0 °F) and the concrete section is less than 300 mm (12 in.)
thick.
In some countries,8.12,8.37 the entire concrete mix is heated to between 40 and 60 °C
(104 and 140 °F). Such temperatures have an adverse effect on workability and on the
long-term strength but these may be balanced by economic considerations: rapid re-
use of formwork and no need for post-placing heating. Also, the high initial
temperature speeds up the process of hydration, so that ‘cost-free’ heat is generated.
Placing against frozen ground should not be permitted and the formwork should, if
possible, be pre-heated.
Following placement, the concrete must be protected from freezing for at least 24
hours. Drying of the surface of the concrete should be prevented, especially when the
concrete is much warmer than the ambient air. However, no active wet curing should
be applied so that the concrete becomes less than saturated. While this may seem to be
counter to the usual recommendations about wet curing, we should note that cold air
(below 10 °C (50 °F)) does not cause excessive drying.
Various types of insulation of concrete placed in cold weather are described in ACI
306R-88 (Reapproved 2007).8.55 What is important is to remove the insulation in a
manner such as to avoid a sudden change in temperature at the surface of the concrete
and the development of steep temperature gradients within the concrete element. ACI
306R-88 also gives information about protecting and heating concrete in cold weather.
It needs to be pointed out that the means of heating should be such that the concrete
does not dry out rapidly, that no part of it is heated excessively, and that no high
concentration of CO2 in the atmosphere results. This last point means that combustion
heaters, unless vented, should not be used in enclosed spaces.
An alternative to placing concrete under conditions such that normal mix water
cannot freeze is to depress the freezing point of the mix water well below 0 °C (32 °F).
This can be achieved by the use of anti-freeze admixtures. Potassium carbonate
(potash) was one of the first such admixtures to be used.8.96 More recent developments
include the use of calcium nitrite and sodium nitrite; it may be recalled that these
inorganic salts act as accelerators (see p. 247) and that they are non-corrosive with
respect to steel. Mixes containing nitrites were found to produce concrete with
significant strengths at temperatures down to –10 °C (14 °F).8.17 As is sometimes the
case with admixtures, anti-freeze admixtures of undisclosed composition are
claimed8.16 to result in air-entrained concrete mixes which gain strength at temperatures
of –7 °C (20 °F) and even down to –19 °C (–2 °F); in the latter case, however, the
solids content of the admixtures is 47 per cent so that the provision of an adequate
amount of mix water may not be possible. Practical acceptance of admixtures of this
type is yet to come.
Without using anti-freeze admixtures, it is possible to place air-entrained concrete
at 0 °C (32 °F) because, as soon as hydration has begun, the freezing point of the pore
water is depressed so that no frost action takes place above about –2 °C (28 °F).
Development of strength of concretes with water/cement ratios of 0.35 and 0.45, cast
at 0 °C (32 °F) and stored in sea water in the laboratory at 0 °C (32 °F), was
determined by Gardner.8.18 He reported long-term strengths, both compressive and
tensile, comparable with those of concrete stored at 16 °C (61 °F). This latter finding
is similar to that of Aïtcin.8.23 Both these investigations indicate that keeping concrete
in sea water at 0 °C (32 °F) is not harmful. This may not be the case with storage in

438
air at the same temperature. In any event, under conditions of natural exposure, the
absence of a drop in temperature below 0 °C (32 °F) cannot be guaranteed.
References
8.1. F. M. LEA and N. DAVEY, The deterioration of concrete in structures, J. Inst.
Civ. Engrs. No. 7, pp. 248–95 (London, May 1949).
8.2. A. NEVILLE, Cement and concrete: their interaction in practice, in Advances in
Cement and Concrete, American Soc. Civil Engineers, pp. 1–14 (New York,
1994).
8.3. N. I. FATTUHI, The setting of mortar mixes subjected to different
temperatures, Cement and Concrete Research, 18, No. 5, pp. 669–73 (1988).
8.4. ACI 207.4R-93, Cooling and insulating systems for mass concrete, ACI
Manual of Concrete Practice, Part 1 – 1992: Materials and General
Properties of Concrete, 22 pp. (Detroit, Michigan, 1994).
8.5. A. J. AL-TAYYIB et al., The effect of thermal cycling on the durability of
concrete made from local materials in the Arabian Gulf countries, Cement
and Concrete Research, 19, No. 1, pp. 131–42 (1989).
8.6. P. SMITH, Resistance to fire and high temperature, in Concrete and Concrete-
Making, Eds P. Klieger and J. F. Lamond, ASTM Sp. Tech. Publ. No. 169C,
pp. 282–95 (Philadelphia, Pa, 1994).
8.7. F. M. LEA, The Chemistry of Cement and Concrete (London, Arnold, 1970).
8.8. U.S. BUREAU OF RECLAMATION, 4909–92, Procedure for thermal diffusivity of
concrete, Concrete Manual, Part 2, 9th Edn, pp. 685–94 (Denver, Colorado,
1992).
8.9. R. SHALON and D. RAVINA, Studies in concreting in hot countries, RILEM Int.
Symp. on Concrete and Reinforced Concrete in Hot Countries (Haifa, July
1960).
8.10. J. M. SCANLON and J. E. MCDONALD, Thermal properties, in Concrete and
Concrete-Making, Eds P. Klieger and J. F. Lamond, ASTM Sp. Tech. Publ.
No. 169C, pp. 299–39 (Philadelphia, Pa, 1994).
8.11. W. H. PRICE, Factors influencing concrete strength, J. Amer. Concr. Inst., 47,
pp. 417–32 (Feb. 1951).
8.12. E. KILPI and H. KUKKO, Properties of hot concrete and its use in winter
concreting, Nordic Concrete Research Publication, No. 1, 11 pp. (1982).
8.13. J. M. SCANLON, Controlling concrete during hot and cold weather, ACI Tuthill
Symposium, ACI SP-104, pp. 241–59 (Detroit, Michigan, 1987).
8.14. ACI 305R-91, Hot weather concreting, ACI Manual of Concrete Practice,
Part 2 – 1992: Construction Practices and Inspection Pavements, 20 pp.
(Detroit, Michigan, 1994).
8.15. H. TAKEUCHI, Y. TSUJI and A. NANNI, Concrete precooling method by means of
dry ice, Concrete International, 15, No. 11, pp. 52–6 (1993).
8.16. J. W. BROOK et al., Cold weather admixture, Concrete International, 10, No.
10, pp. 44–9 (1988).

439
 8.17. C. J. KORHONEN, E. R. CORTEZ and B. A. CHAREST, Strength development of
concrete cured at low temperature, Concrete International, 14, No. 12, pp.
34–9 (1992).
8.18. N. J. GARDNER, P. L. SAU and M. S. CHEUNG, Strength development and
durability of concrete, ACI Materials Journal, 85, No. 6, pp. 529–36 (1988).
8.19. M. KURITA et al., Precooling concrete using frozen sand, Concrete
International, 12, No. 6, pp. 60–5 (1990).
8.20. G. S. HASANAIN, T. A. KAHALLAF and K. MAHMOOD, Water evaporation from
freshly placed concrete surfaces in hot weather, Cement and Concrete
Research, 19, No. 3, pp. 465–75 (1989).
8.21. O. Z. CEBECI, Strength of concrete in warm and dry environment, Materials
and Structures, 20, No. 118, pp. 270–72 (1987).
8.22. M. A. MUSTAFA and K. M. YUSOF, Mechanical properties of hardened concrete
in hot–humid climate, Cement and Concrete Research, 21, No. 4, pp. 601–
13 (1991).
8.23. P-C. AÏTCIN, M. S. CHEUNG and V. K. SHAH, Strength development of concrete
cured under arctic sea conditions, in Temperature Effects on Concrete,
ASTM Sp. Tech. Publ. No. 858, pp. 3–20 (Philadelphia, Pa, 1983).
8.24. M. MITTELACHER, Effect of hot weather conditions on the strength
performance of set-retarded field concrete, in Temperature Effects on
Concrete, ASTM Sp. Tech. Publ. No. 858, pp. 88–106 (Philadelphia, Pa,
1983).
8.25. R. D. GAYNOR, R. C. MEININGER and T. S. KHAN, Effect of temperature and
delivery time on concrete proportions, in Temperature Effects on Concrete,
ASTM Sp. Tech. Publ. No. 858, pp. 68–87 (Philadelphia, Pa, 1983).
8.26. F. PETSCHARNIG, Einflüsse der jahreszeitlichen Temperaturschwankungen auf
die Betondruckfestigkeit, Zement und Beton, 32, No. 4, pp. 162–3 (1987).
8.27. ACI 517.2R-87, Revised 1992, Accelerated curing of concrete at
atmospheric pressure – state of the art, ACI Manual of Concrete Practice
Part 5 – 1992: Masonry, Precast Concrete, Special Processes, 17 pp.
(Detroit, Michigan, 1994).
8.28. Y. DAN, T. CHIKADA and K. NAGAHAMA, Properties of steam cured concrete
used with ground granulated blast-furnace slag, CAJ Proceedings of Cement
and Concrete, No. 45, pp. 222–7 (1991).
8.29. G. P. TOGNON and G. COPPETTI, Concrete fast curing by two-stage low and
high pressure steam cycle, Proceedings International Congress of the
Precast Concrete Industry, Stresa, 15 pp.
8.30. U.S. ARMY CORPS OF ENGINEERS, Test method for coefficient of linear thermal
expansion of concrete, CRD-C 39-81 Handbook for Concrete and Cement, 2
pp. (Vicksburg, Miss., 1981).
8.31. V. DODSON, Concrete Admixtures, 211 pp. (New York, Van Nostrand
Reinhold, 1990).
8.32. C. R. CRUZ and M. GILLEN, Thermal expansion of Portland cement paste,
mortar, and concrete at high temperatures, Fire and Materials, 4, No. 2, pp.
66–70 (1980).

440
 8.33. T. Z. HARMATHY and J. R. MEHAFFEY, Design of buildings for prescribed levels
of structural fire safety, Fire Safety: Science and Engineering, ASTM Sp.
Tech. Publ. No. 882, pp. 160–75 (Philadelphia, Pa, 1985).
8.34. S. D. VENECANIN, Thermal incompatibility of concrete components and
thermal properties of carbonate rocks, ACI Materials Journal, 87, No. 6, pp.
602–7 (1990).
8.35. S. BREDENKAMP, D. KRUGER and G. L. BREDENKAMP, Direct electric curing of
concrete, Mag. Concr. Res., 45, No. 162, pp. 71–4 (1993).
8.36. U. MENZEL, Heat treatment of concrete, Concrete Precasting Plant and
Technology, Issue 12, pp. 92–7 (1991).
8.37. M. MAMILLAN, Traitement thermique des bétons, in Le béton hydraulique, pp.
261–9 (Presses de l’École Nationale des Ponts de Chaussées, Paris, 1982).
8.38. S. A. AUSTIN, P. J. ROBINS and M. R. RICHARDS, Jetblast temperature-resistant
concrete for Harrier aircraft pavements, The Structural Engineer, 79, Nos
23/24, pp. 427–32 (1992).
8.39. M. DIRUY, Variations du coefficient de dilatation et du retrait de dessiccation
des bétons en place dans les ouvrages, Bull. Liaison Laboratoires Ponts et
Chaussés, 186, pp. 45–54 (July–Aug. 1993).
8.40. H. DETTLING, The thermal expansion of hardened cement paste, aggregates,
and concretes, Deutscher Ausschuss für Stahlbeton, Part 2, No. 164, pp. 1–
65 (1964).
8.41. M. Y. L. CHEW, Effect of heat exposure duration on the thermoluminescence
of concrete, ACI Materials Journal, 90, No. 4, pp. 319–22 (1993).
8.42. G. G. CARETTE and V. M. MALHOTRA, Performance of dolostone and limestone
concretes at sustained high temperatures, in Temperature Effects on
Concrete, ASTM Sp. Tech. Publ. No. 858, pp. 38–67 (Philadelphia, Pa,
1983).
8.43. U.-M. JUMPPANEN, Effect of strength on fire behaviour of concrete, Nordic
Concrete Research, Publication No. 8, pp. 116–27 (Oslo, Dec. 1989).
8.44. G. T. G. MOHAMEDBHAI, Effect of exposure time and rates of heating and
cooling on residual strength of heated concrete, Mag. Concr. Res. 38, No.
136, pp. 151–8 (1986).
8.45. G. G. CARETTE, K. E. PAINTER and V. M. MALHOTRA, Sustained high
temperature effect on concretes made with normal portland cement, normal
portland cement and slag, or normal portland cement and fly ash. Concrete
International, 4, No. 7, pp. 41–51 (1982).
8.46. W. P. S. DIAS, G. A. KHOURY and P. J. E. SULLIVAN, Mechanical properties of
hardened cement paste exposed to temperature up to 700 C (1292 F), ACI
Materials Journal, 87, No. 2, pp. 160–6 (1990).
8.47. K. D. HERTZ, Danish investigations on silica fume concrete at elevated
temperatures, ACI Materials Journal, 89, No. 4, pp. 345–7 (1992).
8.48. C. CASTILLO and A. J. DURANNI, Effect of transient high temperature on high-
strength concrete, ACI Materials Journal, 87, No. 1, pp. 47–53 (1990).
8.49. D. BERNER, B. C. GERWICK, JNR and M. POLIVKA, Static and cyclic behavior of
structural lightweight concrete at cryogenic temperatures, in Temperature

441
 Effects on Concrete, ASTM Sp. Tech. Publ. No. 858, pp. 21–37 (Philadelphia,
Pa, 1983).
8.50. T. MIURA, The properties of concrete at very low temperatures, Materials and
Structures, 22, No. 130, pp. 243–54 (1989).
8.51. Y. GOTO and T. MIURA, Experimental studies on properties of concrete cooled
to about minus 160 °C, Technical Reports, Tohoku University, 44, No. 2, pp.
357–85 (1979).
8.52. P.-C. AÏTCIN and N. RIAD, Curing temperature and very high strength
concrete, Concrete International, 10, No. 10, pp. 69–72 (1988).
8.53. B. WILDE, Concrete comments, Concrete International, 15, No. 6, p. 80
(1993).
8.54. ACI 207.1R-87, Mass concrete, ACI Manual of Concrete Practice, Part 1 –
1992: Materials and General Properties of Concrete, 44 pp. (Detroit,
Michigan, 1994).
8.55. ACI 306R-88, Cold weather concreting, ACI Manual of Concrete Practice,
Part 2 – 1992: Construction Practices and Inspection Pavements, 23 pp.
(Detroit, Michigan, 1994).
8.56. K. W. NASSER and M. CHAKRABORTY, Effects on strength and elasticity of
concrete, in Temperature Effects on Concrete, ASTM Sp. Tech. Publ. No.
858, pp. 118–33 (Philadelphia, Pa, 1983).
8.57. T. KANDA, F. SAKURAMOTO and K. SUZUKI, Compressive strength of silica fume
concrete at higher temperatures, in Silica Fume, Slag, and Natural
Pozzolans in Concrete, Vol. II, Ed. V. M. Malhotra, ACI SP-132, pp. 1089–
103 (1992).
8.58. D. N. RICHARDSON, Review of variables that influence measured concrete
compressive strength, Journal of Materials in Civil Engineering, 3, No. 2,
pp. 95–112 (1991).
8.59. A. BENTUR and C. JAEGERMANN, Effect of curing and composition on the
properties of the outer skin of concrete, Journal of Materials in Civil
Engineering, 3, No. 4, pp. 252–62 (1991).
8.60. ACI 308-92, Standard practice for curing concrete, in ACI Manual of
Concrete Practice, Part 2 – 1992: Construction Practices and Inspection
Pavements, 11 pp. (Detroit, Michigan, 1994).
8.61. F. D. BERESFORD and F. A. BLAKEY, Discussion on paper by W. Lerch: Plastic
shrinkage, J. Amer. Concr. Inst., 56, Part II, pp. 1342–3 (Dec. 1957).
8.62. R. SHALON, Report on behaviour of concrete in hot climate, Materials and
Structures, 11, No. 62, pp. 127–31 (1978).
8.63. NATIONAL READY MIXED CONCRETE ASSOCIATION, Cold weather ready mixed
concrete, Publ. No. 34 (Washington DC, Sept. 1960).
8.64. O. BERGE, Improving the properties of hot-mixed concrete using retarding
admixtures. J. Amer. Concr. Inst. 73, pp. 394–8 (July 1976).
8.65. M. E. FITZGIBBON, Large pours for reinforced concrete
structures, Concrete, 10, No. 3, p. 41 (London, March 1976).
8.66. M. E. FITZGIBBON, Large pours – 2, heat generation and control, Concrete, 10,
No. 12, pp. 33–5 (London, Dec. 1976).

442
 8.67. B. MATHER, Use of concrete of low portland cement in combination with
pozzolans and other admixtures in construction of concrete dams. J. Amer.
Concr. Inst., 71, pp. 589–99 (Dec. 1974).
8.68. G. MOLLER, Tests of resistance of concrete to early frost action, RILEM
Symposium on Winter Concreting (Copenhagen, 1956).
8.69. E. G. SWENSON, Winter concreting trends in Europe. J. Amer. Concr. Inst., 54,
pp. 369–84 (Nov. 1957).
8.70. P. KLIEGER, Effect of mixing and curing temperature on concrete strength, J.
Amer. Concr. Inst., 54, pp. 1063–81 (June 1958).
8.71. U.S. BUREAU OF RECLAMATION, Concrete Manual, 8th Edn (Denver, Colorado,
1975).
8.72. A. G. A. SAUL, Steam curing and its effect upon mix design, Proc. of a
Symposium on Mix Design and Quality Control of Concrete, pp. 132–42
(London, Cement and Concrete Assoc., 1954).
8.73. J. J. SHIDELER and W. H. CHAMBERLIN, Early strength of concretes as affected
by steam curing temperatures, J. Amer. Concr. Inst., 46, pp. 273–82 (Dec.
1949).
8.74. K. O. KJELLSEN, R. J. DETWILER and O. E. GJØRV, Backscattered electron
imaging of cement pastes hydrated at different temperatures, Cement and
Concrete Research, 20, No. 2, pp. 308–11 (1990).
8.75. H. F. GONNERMAN, Annotated Bibliography on High-pressure Steam Curing of
Concrete and Related Subjects (National Concrete Masonry Assoc., Chicago,
1954).
8.76. T. THORVALDSON, Effect of chemical nature of aggregate on strength of steam-
cured portland cement mortars, J. Amer. Concr. Inst., 52, pp. 771–80 (1956).
8.77. G. J. VERBECK and R. A. HELMUTH, Structures and physical properties of
cement paste, Proc. 5th Int. Symp. on the Chemistry of Cement, Tokyo, Vol.
3, pp. 1–32 (1968).
8.78. C. N. NAGARAJ and A. K. SINHA, Heat-resisting concrete, Indian Concrete
J., 48, No. 4, pp. 132–7 (April 1974).
8.79. J. A. HANSON, Optimum steam curing procedures for structural lightweight
concrete, J. Amer. Concr. Inst., 62, pp. 661–72 (June 1965).
8.80. B. D. BARNES, R. L. ORNDORFF and J. E. ROTEN, Low initial curing temperature
improves the strength of concrete test cylinders. J. Amer. Concr. Inst., 74,
No. 12, pp. 612–15 (1977).
8.81. CEMENT AND CONCRETE ASSOCIATION, Research and development – Research on
materials. Annual Report, pp. 14–19 (Slough, 1976).
8.82. J. ALEXANDERSON, Strength loss in heat curing – causes and
countermeasures, Behavior of Concrete under Temperature Extremes, ACI
SP-39, pp. 91–107 (Detroit, Michigan, 1973).
8.83. I. SOROKA, C. H. JAEGERMANN and A. BENTUR, Short-term steam-curing and
concrete later-age strength, Materials and Structures, 11, No. 62, pp. 93–6
(1978).

443
 8.84. G. VERBECK and L. E. COPELAND, Some physical and chemical aspects of high-
pressure steam curing, Menzel Symposium on High-Pressure Steam Curing,
ACI SP-32, pp. 1–13 (Detroit, Michigan, 1972).
8.85. C. J. DODSON and K. S. RAJAGOPALAN, Field tests verify temperature effects on
concrete strength, Concrete International, 1, No. 12, pp. 26–30 (1979).
8.86. R. SUGIKI, Accelerated hardening of concrete (in Japanese), Concrete
Journal, 12, No. 8, pp. 1–14 (1974).
8.87. N. DAVEY, Concrete mixes for various building purposes, Proc. of a
Symposium on Mix Design and Quality Control of Concrete, pp. 28–41
(London, Cement and Concrete Assn, 1954).
8.88. S. L. MEYERS, HOW temperature and moisture changes may affect the
durability of concrete. Rock Products, pp. 153–7 (Chicago, Aug. 1951).
8.89. S. WALKER, D. L. BLOEM and W. G. MULLEN, Effects of temperature changes
on concrete as influenced by aggregates, J. Amer. Concr. Inst., 48, pp. 661–
79 (April 1952).
8.90. D. G. R. BONNELL and F. C. HARPER, The thermal expansion of
concrete, National Building Studies, Technical Paper No. 7 (London,
HMSO, 1951).
8.91. T. C. POWERS and T. L. BROWNYARD, Studies of the physical properties of
hardened portland cement paste (Nine parts), J. Amer. Concr. Inst., 43 (Oct.
1946 to April 1947).
8.92. R. PHILLEO, Some physical properties of concrete at high temperatures, J.
Amer Concr. Inst., 54, pp. 857–64 (April 1958).
8.93. N. G. ZOLDNERS, Effect of high temperatures on concretes incorporating
different aggregates, Mines Branch Research Report R.64, Department of
Mines and Technical Surveys (Ottawa, May 1960).
8.94. S. L. MEYERS, Thermal coefficient of expansion of portland cement – Long-
time tests, Industrial and Engineering Chemistry, 32, No. 8, pp. 1107–12
(Easton, Pa, 1940).
8.95. H. L. MALHOTRA, The effect of temperature on the compressive strength of
concrete, Mag. Concr. Res., 8, No. 23, pp. 85–94 (1956).
8.96. M. G. DAVIDSON, A New Cold Weather Concrete Technology (Potash as a
Frost-resistant Admixture) (Moscow, Lenizdat, 1966).
8.97. A. G. LOUDON and E. F. STACEY, The thermal and acoustic properties of
lightweight concretes, Structural Concrete, 3, No. 2, pp. 58–95 (London,
1966).
8.98. T. HARADA, J. TAKEDA, S. YAMANE and F. FURUMURA, Strength, elasticity and the
thermal properties of concrete subjected to elevated temperatures, Int.
Seminar on Concrete for Nuclear Reactors, ACI SP-34, 1, pp. 377–406
(Detroit, Michigan, 1972).
8.99. H. W. BREWER, General relation of heat flow factors to the unit weight of
concrete, J. Portl. Cem. Assoc. Research and Development Laboratories, 9,
No. 1, pp. 48–60 (Jan. 1967).

444
 8.100. R. A. HELMUTH, Dimensional changes of hardened portland cement pastes
caused by temperature changes, Proc. Highw. Res. Board, 40, pp. 315–36
(1961).
8.101. D. J. HANNANT, Effects of heat on concrete strength, Engineering, 197, p. 302
(London, Feb. 21, 1964).
8.102. K. W. NASSER and A. M. NEVILLE, Creep of concrete at elevated
temperatures, J. Amer. Concr. Inst., 62, pp. 1567–79 (Dec. 1965).
8.103. M. S. ABRAMS and A. H. GUSTAFERRO, Fire endurance of concrete slabs as
influenced by thickness, aggregate type, and moisture, J. Portl. Cem. Assoc.
Research and Development Laboratories, 10, No. 2, pp. 9–24 (May 1968).
8.104. J. C. MARÉCHAL, Variations in the modulus of elasticity and Poisson’s ratio
with temperature, Int. Seminar on Concrete for Nuclear Reactors, ACI SP-
34, 1, pp. 495–503 (Detroit, Michigan, 1972).
8.105. RILEM WINTER CONSTRUCTION COMMITTEE, Recommandations pour le bétonnage
en hiver, Supplément aux Annales de l’Institut Technique du Bâtiment et des
Travaux Publics, No. 190, Béton, Béton Armé No. 72, pp. 1012–37 (Oct.
1963).
8.106. U. TRÜB, Baustoff Beton (Wildegg, Switzerland, Technische Forschungs und
Beratungsstelle der Schweizerischen Zementindustrie, 1968).
8.107. F. WITTMANN and J. LUKAS, Experimental study of thermal expansion of
hardened cement paste, Materials and Structures, 7, No. 40, pp. 247–52
(1974).
8.108. M. S. ABRAMS, Compressive strength of concrete at temperatures to
1600F, Temperature and Concrete, ACI SP-25, pp. 33–58 (Detroit,
Michigan, 1971).
8.109. L. JOHANSSON, Flame cleaning of concrete, CBI Reports, 15:75, 6 pp.
(Swedish Cement and Concrete Research Inst., 1975).
8.110. D. WHITING, A. LITVIN and S. E. GOODWIN, Specific heat of selected
concretes, J. Amer. Concr. Inst., 75, No. 7, pp. 299–305 (1978).
8.111. D. R. LANKARD, D. L. BIRKIMER, F. F. FONDRIEST and M. J. SNYDER, Effects of
moisture content on the structural properties of portland cement concrete
exposed to temperatures up to 500F, Temperature and Concrete, ACI SP-25,
pp. 59–102 (Detroit, Michigan, 1971).
8.112. R. SARSHAR and G. A. KHOURY, Material and environmental factors
influencing the compressive strength of unsealed cement paste and concrete
at high temperatures, Mag. Concr. Res., 45, No. 162, pp. 51–61 (1993).
8.113. S. GOTO and D. M. ROY, The effect of w/c ratio and curing temperature on the
permeability of hardened cement paste, Cement and Concrete Research, 11,
No. 4, pp. 575–9 (1981).
8.114. L. KRISTENSEN and T. C. HANSEN, Cracks in concrete core due to fire or thermal
heating shock, ACI Materials Journal, 91, No. 5, pp. 453–9 (1994).
8.115. A. NEVILLE, Neville on Concrete: An Examination of Issues in Concrete
Practice, Second Edition (Book Surge LLC, www.createspace.com, 2006).

445
 8.116. E. MENÉNDEZ and L. VEGA, Analysis of behaviour of the structural concrete
after the fire at the Windsor Building in Madrid, Fire and Materials, 34, pp.
95–107 (2009).
8.117. P. J. E. SULLIVAN, A probabilistic method of testing for the assessment of
deterioration and explosive spalling of high strength concrete beams in
flexure at high temperature, Cement and Concrete Composites, 26, pp. 155–
162 (2004).

446
Chapter 9. Elasticity, shrinkage, and creep

Much of the discussion in the preceding chapters referred to the strength of concrete,
which is of significant importance in the design of concrete structures. However, with
any stress, there is always associated a strain, and vice versa. Strain can arise also
from causes other than applied stress. The relation between stress and strain over their
full range is of vital interest in structural design. The topic of strain and, more
generally, of the different types of deformation of concrete is the subject matter of this
chapter.
Like many other structural materials, concrete is, to a certain degree, elastic. A
material is said to be perfectly elastic if strain appears and disappears immediately on
application and removal of stress. This definition does not imply a linear stress–strain
relation: elastic behaviour coupled with a non-linear stress–strain relation is exhibited,
for instance, by glass and some rocks.
When concrete is subjected to sustained loading, strain increases with time, i.e.
concrete exhibits creep. In addition, whether subjected to load or not, concrete
contracts on drying, undergoing shrinkage. The magnitudes of shrinkage and creep
are of the same order as elastic strain under the usual range of stresses, so that the
various types of strain must be, at all times, taken into account.
Stress–strain relation and modulus of elasticity
Figure 9.1 shows a diagrammatic representation of the stress–strain relation for a
concrete specimen loaded and unloaded in compression or tension up to a stress well
below the ultimate strength. In compression tests, a small concave-up part of the
curve at the beginning of loading is sometimes encountered; this is due to the closing
of pre-existing fine shrinkage cracks. From Fig. 9.1, it can be seen that the term
Young’s modulus of elasticity can, strictly speaking, be applied only to the straight
part of the stress–strain curve, or, when no straight portion is present, to the tangent to
the curve at the origin. This is the initial tangent modulus, but it is of limited practical
importance. It is possible to find a tangent modulus at any point on the stress–strain
curve, but this modulus applies only to very small changes in load above or below the
load at which the tangent modulus is considered.

447
 Fig. 9.1. Diagrammatic representation of the stress–strain relation for concrete
The magnitude of the observed strains and the curvature of the stress–strain
relation depend, at least in part, on the rate of application of stress. When the load is
applied extremely rapidly, say, in less than 0.01 second, recorded strains are greatly
reduced, and the curvature of the stress–strain curve becomes extremely small. An
increase in loading time from 5 seconds to about 2 minutes can increase the strain by
up to 15 per cent, but within the range of 2 to 10 (or even 20) minutes – a time
normally required to test a specimen in an ordinary testing machine – the increase in
strain is very small. The relation between the rate of strain and strength, discussed on
p. 621, may be of relevance.
The increase in strain while the load, or part of it, is acting is due to creep of
concrete, but the dependence of instantaneous strain on the speed of loading makes
the demarcation between elastic and creep strains difficult. For practical purposes, an
arbitrary distinction is made: the deformation occurring during loading is considered
elastic, and the subsequent increase in strain is regarded as creep. The modulus of
elasticity satisfying this requirement is the secant modulus of Fig. 9.1, also known as
the chord modulus. The secant modulus is a static modulus because it is determined
from an experimental stress–strain relation on a test cylinder, in contradistinction to
the dynamic modulus, considered on p. 421.
Because the secant modulus decreases with an increase in stress, the stress at
which the modulus has been determined must always be stated. For comparative
purposes, the maximum stress applied is chosen as a fixed proportion of the ultimate
strength. This proportion is prescribed as 33 per cent in BS 1881-121 : 1983, and as
40 per cent in ASTM C 469-02. To eliminate creep, and also to achieve seating of the
gauges, at least two cycles of pre-loading to the maximum stress are required. The

448
minimum stress must be such that the test cylinder does not move. This minimum is
specified by BS 1881-121 : 1983 as 0.5 MPa; ASTM C 469-02 specifies a minimum
strain of 50 × 10–6. The stress–strain curve on the third or fourth loading exhibits only
a small curvature.
It is interesting to note that the two components of concrete, that is, hydrated
cement paste and aggregate, when individually subjected to load, exhibit a sensibly
linear stress–strain relation (Fig. 9.2), although some suggestions about the non-
linearity of the stress–strain relation of the hydrated cement paste have been
made.9.100 The reason for the curved relation in the composite material – concrete – lies
in the presence of interfaces between the cement paste and the aggregate and in the
development of bond microcracks at those interfaces.9.42 The progressive development
of microcracking was confirmed by neutron radiography.9.62

Fig. 9.2. Stress–strain relations for cement paste, aggregate, and concrete
The development of microcracking means that the stored strain energy is
transformed into the surface energy of the new crack faces. Because the cracks
develop progressively at interfaces making varying angles with the applied load, and
respond to the local stress, there is a progressive increase in local stress intensity and
in the magnitude of strain. In other words, a consequence of the development of the
cracks is a reduction in the effective area resisting the applied load, so that the local
stress is larger than the nominal stress based on the total cross-section of the specimen.
These changes mean that the strain increases at a faster rate than the nominal applied

449
stress, and so the stress–strain curve continues to bend over, with an apparent pseudo-
plastic behaviour.9.43
When the applied stress increases beyond approximately 70 per cent of the
ultimate strength, mortar cracking (connecting the bond cracks) develops (see p. 300)
and the stress–strain curve bends over at an increasing rate. The development of a
continuous crack system reduces the number of load-carrying paths9.65 and, eventually,
the ultimate strength of the specimen is reached. This is the peak of the stress–strain
curve.
If the testing machine allows a reduction in the applied load, the strain will
continue to increase with a decrease in the nominal applied stress. This is the post-
peak part of the stress–strain curve which represents strain softening of concrete.
However, the observed descending part of the stress–strain curve is not
a material property9.65 but is affected by test conditions. The main influencing factors
are the stiffness of the testing machine in relation to the stiffness of the test specimen
and the rate of strain.9.67 A typical complete stress–strain curve is shown in Fig. 9.3.9.36

Fig. 9.3. Stress–strain relation of concretes tested in compression at a constant

rate of strain9.36
It can be noted that, if the stress–strain curve ended abruptly at the peak, the
material would be classified as brittle. The less steep the descending part of the
stress–strain curve the more ductile the behaviour. If the slope beyond the peak were
zero, the material would be said to be perfectly plastic.
In structural design of reinforced concrete, the entire stress–strain curve, often in
idealized form, must be considered. For this reason, the behaviour of concrete which
has a very high strength is of especial interest. Such concrete develops a smaller
amount of cracking than normal-strength concrete during all stages of loading;9.66 in
consequence, the ascending part of the stress–strain curve is steeper and linear up to a

450
very high proportion of the ultimate strength. The descending part of the curve is also
very steep (see Fig. 9.4) so that high strength concrete is more brittle than ordinary
concrete, and indeed explosive failure of a local part of specimens of high strength
concrete tested in compression has often been encountered. However, the apparent
brittleness of high strength concrete is not necessarily reflected in the behaviour of
reinforced concrete members made with such concrete.9.63,9.64

Fig. 9.4. Examples of stress–strain relation in compression for concrete cylinders

with compressive strength up to 85 MPa
The behaviour of high-strength concrete is of interest also with respect to the strain
at various stress levels. If the stress considered, for example the stress in service, is
expressed as a fraction of the ultimate strength, referred to as the stress/strength ratio,
then the following observations can be made. At the same stress/strength ratio, the
stronger the concrete the larger the strain. At the maximum stress, that is at the stress
corresponding to the ultimate strength, in a 100 MPa (15 000 psi) concrete, the strain
is typically 3 × 10–3 to 4 × 10–3; in a 20 MPa (3000 psi) concrete, the strain is about 2 ×

451
10–3. However, under the same stress, regardless of strength, stronger concrete exhibits
a lower strain. It follows that high-strength concrete has a higher modulus of elasticity,
as seen in Fig. 9.4.
Parenthetically, we can observe that this behaviour is in contrast to that of different
grades of steel, possibly because the strength of hydrated cement paste is governed by
the gel/space ratio, which can be expected to affect also the stiffness of the
cementitious material. On the other hand, the strength of steel is related to the
structure and boundaries of crystals but not to voids, so that the stiffness of the
material is unaffected by its strength.
Lightweight aggregate concrete exhibits a steeper descending part of the stress–
strain curve,9.36 (see Fig. 9.3) that is, it has a somewhat more brittle behaviour than
normal weight concrete.
The stress–strain curve in tension is similar in shape to that in compression
(see Fig. 9.5) but a special testing machine is necessary.9.61 In direct tension, the
development of cracks has the effect both of reducing the effective area resisting
stress and of increasing the contribution of cracks to the overall strain. This may be
the reason why the departure from linearity of the stress–strain relation in tension
occurs at a slightly lower stress/strength ratio than in compression.9.34

Fig. 9.5. Examples of the stress–strain relation in direct tension (based on

ref. 9.61)
Expressions for stress–strain curve
Because the precise shape of the entire stress–strain curve for concrete is not a
property of the material per se but depends on the test arrangements, there is little of
fundamental importance in formulating an equation for the stress–strain relation. This
is not to deny the usefulness of such an equation in structural analysis. Numerous

452
attempts to develop equations have been made, but probably the most successful
equation was suggested by Desayi and Krishnan:9.44

where ε = strain
σ = stress
ε0 = strain at maximum stress, and
E = initial tangent modulus, assumed to be twice the secant modulus at
maximum stress σmax, i.e.

.
The last assumption is questionable because both σmax and ε0 are strongly affected by
test conditions, and a more general form of the equation which is not constrained by
this assumption has been developed by Carreira and Chu.9.67
Expressions for modulus of elasticity
There is no doubt that the modulus of elasticity increases with an increase in the
compressive strength of concrete, but there is no agreement on the precise form of the
relationship. This is not surprising, given the fact that the modulus of elasticity of
concrete is affected by the modulus of elasticity of the aggregate and by the
volumetric proportion of aggregate in the concrete. The former is rarely known so that
some expressions, for example that of ACI 318-029.98 allow for the modulus of
elasticity of aggregate by a coefficient which is a function of the density of the
concrete, usually density raised to power 1.5.
All that can be said reliably is that the increase in the modulus of elasticity of
concrete is progressively lower than the increase in compressive strength. According
to ACI 318-029.98 the modulus is proportional to the strength raised to power 0.5. The
expression for the secant modulus of elasticity of concrete, Ec, in pounds per square
inch, recommended by ACI 318-029.98 for structural calculations, applicable to normal
weight concrete, is

Ec = 57 000( )0.5

where is the compressive strength of standard test cylinders in pounds per square
inch. When Ec is expressed in GPa and in MPa, the expression is

Ec = 4.73( )0.5.

Some other expressions use the power index of 0.33, instead of 0.5, and also add a
constant term to the right-hand side of the equation.
For concretes with strengths up to 83 MPa (12 000 psi) ACI 363R-929.99 quotes

Ec = 3.32( )0.5 + 6.9

453
where Ec is expressed in GPa and in MPa. In the range of strengths between 80
and 140 MPa (12 000 and 20 000 psi) Kakizaki et al.9.95 found that the modulus of
elasticity, Ec, is approximately related to strength , by the expression

Ec = 3.65( )0.5

using the same units of measurement as above. The modulus was found not to be
affected by curing but was influenced by the modulus of elasticity of the coarse
aggregate in the concrete. This dependence is a consequence of the two-phase nature
of concrete.9.84 The quality of the bond between the two phases is of importance and
may affect the value of the modulus of elasticity of concrete when the bond is
particularly strong, as is the case in high performance concrete (see p. 678).
Furthermore, because such concrete is made with high strength aggregate, which is
likely to have a high modulus of elasticity, high performance concrete tends to have a
higher modulus of elasticity than would be expected from an extrapolation of
expressions for ordinary concrete.
When the density of concrete, ρ, is between 145 and 155 lb/ft3 (taken to be the
range for normal weight concrete) and is expressed in pounds per cubic foot, the
modulus of elasticity is given by ACI 318-029.98 as

Ec = 33ρ1.5( )0.5.

In SI units, this expression becomes

Ec = 43ρ1.5( )0.5 × l0–6.

The use of the power coefficient of 1.5 applied to the density of concrete may not
be correct. According to Lydon and Balendran,9.70 the modulus of elasticity of
aggregate is proportional to the square of its density. Whatever the value of the power
index, the argument is that, at a constant aggregate content, the density of concrete
increases with the increase in the density of aggregate.
The two-phase nature of concrete also means that the volumetric proportions of
aggregate and of hydrated cement paste affect the value of the modulus of elasticity at
a given strength of concrete. Because normal weight aggregate has a higher modulus
of elasticity than hydrated cement paste, a higher content of a given aggregate results
in a higher modulus of elasticity of concrete of a given compressive strength.
Lightweight aggregate has a lower density than hydrated cement paste and
influences the modulus of elasticity of concrete accordingly. The consideration of
density of the concrete in the expression of ACI 318-029.98 means that lightweight
aggregate concrete can be covered by the same expression. We can note that, because
the modulus of elasticity of lightweight aggregate differs little from the modulus of
the hardened cement paste, mix proportions do not affect the modulus of elasticity of
lightweight aggregate concrete.9.7
For concrete cast and stored at 0 °C (32 °F) the rate of change in the modulus of
elasticity with an increase in the strength of concrete was found to be somewhat
steeper than at room temperature9.59 but the difference does not appear to be of
importance.
So far we have considered the modulus of elasticity in compression. Few data are
available for the modulus of elasticity of concrete in tension, which can be determined

454
in direct tension or from measurement of deflection of flexure specimens; where
necessary, a correction for shear should be applied.9.5 The best assumption which can
be made about the modulus of elasticity in tension is that it is equal to the modulus in
compression. This was broadly established by tests,9.34,9.70 and can also be seen from a
comparison of Figs 9.4 and 9.5.
The modulus of elasticity in shear (modulus of rigidity) is not normally determined
by direct measurement.
Curing conditions per se are not believed to affect the modulus of elasticity other
than through the influence of curing on strength. Some reports to the contrary9.69 may
possibly be explained by the fact that the strength of standard test specimens was
considered rather than the strength of the actual concrete. Furthermore, it is necessary
to distinguish between the influence of curing on the modulus of elasticity, which also
affects strength, on the one hand and, on the other, the influence of the moisture
condition during the test. The effects of the latter on the modulus of elasticity and on
strength need not be the same; this is discussed on p. 602.
Dynamic modulus of elasticity
The preceding section dealt exclusively with the static modulus of elasticity, which
gives the strain response to an applied stress of known intensity. There exists another
type of modulus, known as the dynamic modulus, which is determined by means of
vibration of a concrete specimen, only a negligible stress being applied. The
procedure for determining the dynamic modulus of elasticity is described on p. 636.
Because of the absence of a significant applied stress, no microcracking is induced
in the concrete and there is no creep. In consequence, the dynamic modulus refers to
almost purely elastic effects. For this reason, the dynamic modulus is considered to be
approximately equal to the initial tangent modulus determined in the static test and is,
therefore, appreciably higher than the secant modulus which is determined by
application of load to a concrete specimen. This view has, however, been
challenged,9.68 and it has to be recognized that the heterogeneity of concrete affects the
two moduli in different ways.9.1 It cannot, therefore, be expected that there exists
between the two moduli a single relation, based on physical behaviour.
The ratio of the static modulus of elasticity to the dynamic modulus, which is
always smaller than unity, is higher the higher the strength of concrete9.9 and, probably
for this reason, increases with age.9.1 This variable ratio of the moduli means that there
is no simple conversion of the value of the dynamic modulus, Ed, which is easy to
determine, into an estimate of the static modulus, Ec, the knowledge of which is
required in structural design. Nevertheless, various empirical relations, valid over a
limited range, have been developed. The simplest of these, proposed by Lydon and
Balendran,9.70 is

Ec = 0.83Ed.

An expression which used to be included in the British code for design of concrete
structures, BS CP 110:1972, is

Ec = 1.25Ed – 19

both moduli being expressed in GPa. This expression does not apply to concretes
containing more than 500 kg of cement per cubic metre of concrete (850 lb/yd3) or to

455
lightweight aggregate concrete. For the latter, the following expression was
suggested:9.39

Ec = 1.04Ed – 4.1.

For both lightweight and normal concretes, Popovics9.57 suggested that the relation
between the static and dynamic moduli is a function of density of the concrete, just as
is the case with the relation between the static modulus and strength, namely,

where ρ is the density of concrete and k is a constant dependent on the units of

measurement.
Whatever the relation between the moduli, it is thought to be unaffected by air
entrainment, method of curing, condition at test, or the type of cement used.9.11
The dynamic modulus of elasticity is of considerable value in studying changes in
a single test specimen, for example, in consequence of chemical attack.
Poisson’s ratio
When a uniaxial load is applied to a concrete specimen it produces a longitudinal
strain in the direction of the applied load and, at the same time, a lateral strain of
opposite sign. The ratio of the lateral strain to the longitudinal strain is called
Poisson’s ratio; the sign of the ratio is ignored. Normally, we are interested in the
consequences of an applied compression so that the lateral strain is tensile, but the
situation is analogous when a tensile load is applied.
For an isotropic and linear-elastic material, Poisson’s ratio is constant but, in
concrete, Poisson’s ratio may be influenced by specific conditions. However, for
stresses for which the relation between the applied stress and the longitudinal strain is
linear, the value of Poisson’s ratio for concrete is approximately constant. Depending
on the properties of the aggregate used, Poisson’s ratio of concrete lies generally in
the range of 0.15 to 0.22 when determined from strain measurements under a
compressive load. The value of Poisson’s ratio under tensile load appears to be the
same as in compression.9.70
No systematic data on the influence of various factors on Poisson’s ratio are
available. Lightweight aggregate concrete has been reported to have Poisson’s ratio at
the lower end of the range.9.70 The value of Poisson’s ratio has been reported9.94 not to
be affected by the increase in strength with age or by the richness of the mix. The
latter observation needs to be confirmed because the elastic properties of coarse
aggregate can be expected to influence the elastic behaviour of concrete. No
generalizations about Poisson’s ratio can, therefore, be made, but this lack of
information is not critical in view of the fact that, for the majority of concretes, the
range of values is small: 0.17 to 0.20.
Tests on saturated mortar have shown that the value of Poisson’s ratio is higher at
higher rates of strain; for instance, it was found9.60 to increase from 0.20 at a strain rate
of 3 × 10–6 per second to 0.27 at a strain rate of 0.15 per second. This effect may not
be generally valid.
Figure 9.6 shows a typical plot of longitudinal strain and lateral strain under a
steadily and rapidly increasing axial compressive load applied to a cylinder. In
addition, the volumetric strain is plotted. It can be seen that, above a certain stress,
Poisson’s ratio increases rapidly; this is caused by extensive vertical cracking so that,

456
in fact, we are dealing with an apparent Poisson’s ratio. Under a further increase in
stress, the rate of change in volumetric strain changes sign; further on, Poisson’s ratio
exceeds the value of 0.5, and the volumetric strain becomes tensile. The concrete is no
longer a truly continuous body; this is the stage of collapse (cf. p. 294).

Fig. 9.6. Longitudinal, lateral, and volumetric strains in a concrete cylinder

under an increasing stress
It is possible also to determine Poisson’s ratio dynamically. The physical situation
in such a test is distinct from that under static loading as is the case in the
determination of the dynamic modulus of elasticity (see p. 634). For this reason, the
value of Poisson’s ratio determined dynamically is higher than obtained from static
tests; an average value9.5 is about 0.24.
The dynamic method of determining Poisson’s ratio requires the measurement of
pulse velocity, V, and also the fundamental resonant frequency, n, of longitudinal
vibration of a beam of length L (see p. 636). Poisson’s ratio μ can then be calculated
from the expression9.12

since Ed/ρ = (2nL)2, where ρ = density of concrete.

Poisson’s ratio can also be found from the dynamic modulus of elasticity Ed, as
determined in longitudinal or transverse mode of vibration (see p. 634), and the
modulus of rigidity G, using the expression

457
The value of G is normally determined from the resonant frequency of torsional
vibration (see p. 634). Values of μ obtained by this method are intermediate between
those from direct static measurements and those from dynamic tests.
Under sustained stress, the ratio of the lateral strain to the longitudinal strain can
be called creep Poisson’s ratio. Data on this ratio are scarce. At low stresses, creep
Poisson’s ratio is unaffected by the level of stress, indicating that the longitudinal and
lateral deformations due to creep are in the same ratio as the corresponding elastic
deformations. This means that the volume of concrete decreases with the progress of
creep. Above a stress/strength ratio of about 0.5, creep Poisson’s ratio increases
significantly and progressively with the increase in sustained stress.9.93 At a sustained
stress/strength ratio in excess of 0.8 to 0.9, creep Poisson’s ratio exceeds 0.5 and, with
time under sustained stress, failure occurs9.102 (see p. 456).
Under sustained multiaxial compression,9.45 the creep Poisson’s ratio is smaller:
0.09 to 0.17.
Early volume changes
When water moves out of a porous body which is not fully rigid, contraction takes
place. In concrete, from its fresh state to later in life, such movement of water
generally occurs. The various stages of water movement and its consequences will
now be considered.
When discussing the progress of hydration of cement, the resulting changes in
volume were mentioned. The chief of these is the reduction in the volume of the
system cement-plus-water: while the cement paste is plastic, it undergoes a volumetric
contraction whose magnitude is of the order of one per cent of the absolute volume of
dry cement.9.13 However, the extent of hydration prior to setting is small and, once a
certain rigidity of the system of hydrating cement paste has developed, the contraction
induced by the loss of water by hydration is greatly restrained.
Water can also be lost by evaporation from the surface of the concrete while it is
still in the plastic state. A similar loss can occur by suction by the underlying dry
concrete or soil.9.14 This contraction is known as plastic shrinkage because the concrete
is still in the plastic state. The magnitude of plastic shrinkage is affected by the
amount of water lost from the surface of the concrete, which is influenced by
temperature, ambient relative humidity, and wind velocity (see Table 9.1). However,
the rate of loss of water per se does not predict plastic shrinkage;9.103 much depends on
the rigidity of the mix. If the amount of water lost per unit area exceeds the amount of
water brought to the surface by bleeding (see p. 207) and is large, surface cracking
can occur. This is known as plastic shrinkage cracking, referred to on p. 400.
Complete prevention of evaporation immediately after casting eliminates cracking.9.47

Table 9.1. Plastic Shrinkage of Neat Cement Paste Stored in Air at a Relative
Humidity of 50 per cent and Temperature of 20 °C (68 °F)9.14

458
 As stated in the section on Concreting in Hot Weather, the effective means of
preventing plastic shrinkage cracking is by keeping down the rate of evaporation of
water from the surface of concrete: it is recommended that the value of 1 kg/m2 per
hour (0.2 lb/ft2 per hour) should not be exceeded.9.97 It should be remembered that
evaporation is increased when the temperature of the concrete is much higher than the
ambient temperature; under such circumstances, plastic shrinkage can occur even if
the relative humidity of the air is high. It is, therefore, best to protect the concrete
from sun and wind, to place and finish fast, and to start curing very soon thereafter.
Placing concrete on a dry subgrade should be avoided.
Cracking develops also over obstructions to uniform settlement, e.g. reinforcement
or large aggregate particles; this is plastic settlement cracking, which is discussed in
the section on Concreting in Hot Weather. Plastic cracking can develop also when a
large horizontal area of concrete makes contraction in the horizontal direction more
difficult than vertically: deep cracks of an irregular pattern are then formed.9.15 Such
cracks can properly be called pre-setting cracks. Typical plastic shrinkage cracks are
usually parallel to one another, spaced 0.3 to 1 m (1 to 3 ft) apart, and are of
considerable depth. They do not normally extend to free edges of the concrete because
unrestrained contraction is possible there.
Plastic shrinkage is greater the greater the cement content of the mix9.14 (Fig. 9.7)
and the lower the water/cement ratio.9.73 The relation between bleeding and plastic
shrinkage is not straightforward;9.15 for example, retardation of setting allows more
bleeding and leads to increased plastic shrinkage.9.73 On the other hand, greater
bleeding capacity prevents too rapid a drying out of the surface of the concrete and
this reduces plastic shrinkage cracking. In practice, it is the cracking that matters.

459
Fig. 9.7. Influence of cement content of the mix on early shrinkage in air at 20 °C
(68 °F) and 50 per cent relative humidity with wind velocity of 1.0 m/s (2.25
mph)9.14
Autogenous shrinkage
Volume changes occur also after setting has taken place, and may be in the form of
shrinkage or swelling. Continued hydration, when a supply of water is present, leads
to expansion (see next section) but, when no moisture movement to or from the
cement paste is permitted, shrinkage occurs. This shrinkage is the consequence of
withdrawal of water from the capillary pores by the hydration of the hitherto
unhydrated cement, a process known as self-desiccation.
Shrinkage of such a conservative system is known as autogenous shrinkage
or autogenous volume change, and it occurs in practice in the interior of a concrete
mass. The contraction of the cement paste is restrained by the rigid skeleton of the
already hydrated cement paste (mentioned in the preceding section) and also by the
aggregate particles. In consequence, autogenous shrinkage of concrete is an order of
magnitude smaller than in neat cement paste.9.74
Although autogenous shrinkage is three-dimensional, it is usually expressed as a
linear strain so that it can be considered alongside the drying shrinkage. Typical
values of autogenous shrinkage are about 40 × 10–6 at the age of one month and 100 ×

460
10–6 after five years.9.17 Autogenous shrinkage tends to increase at higher temperatures,
with a higher cement content, and possibly with finer cements,9.46 and with cements
which have a high C3A and C4AF content. At a constant content of blended cement, a
higher content of fly ash leads to lower autogenous shrinkage.9.46 As self-desiccation is
greater at lower water/cement ratios, autogenous shrinkage could be expected to
increase but this may not occur because of the more rigid structure of the hydrated
cement paste at low water/cement ratios. Nevertheless, at very low water/cement
ratios, autogenous shrinkage is very high: a value of 700 × 10–6 was reported9.88 for
concrete with a water/cement ratio of 0.17.
As stated in the preceding paragraph, autogenous shrinkage is relatively small,
except at extremely low water/cement ratios autogenous shrinkage. For practical
purposes (other than in large mass concrete structures) such as used in high
performance concrete need not be distinguished from shrinkage caused by drying out
of concrete. The latter is known as drying shrinkage and, in practice, normally
includes that contraction which is due to autogenous volume change.
An integrated view of the various types of shrinkage is given in ref. 9.159.
Swelling
Cement paste or concrete cured continuously in water from the time of placing
exhibits a net increase in volume and an increase in mass. This swelling is due to the
absorption of water by the cement gel: the water molecules act against the cohesive
forces and tend to force the gel particles further apart, with a resultant swelling
pressure. In addition, the ingress of water decreases the surface tension of the gel, and
a further small expansion takes place.9.18
Linear expansion of neat cement paste (relative to the dimensions 24 hours after
casting) has typical values of:9.14
1300 × 10–6 after 100 days
2000 × 10–6 after 1000 days, and
2200 × 10–6 after 2000 days.
These values of swelling, like those of shrinkage and creep, are expressed as linear
strain in metres per metre or inches per inch.
The swelling of concrete is considerably smaller, approximately 100 × 10–6 to 150
× 10–6 for a mix with a cement content of 300 kg/m3 (500 lb/yd3).9.14 This value is
reached 6 to 12 months after casting, and only a very small further swelling takes
place.
Swelling is accompanied by an increase in mass of the order of 1 per cent.9.14 The
increase in mass is thus considerably greater than the increase in volume because
water enters to occupy the space created by the decrease in volume on hydration of
the system cement-plus-water.
Swelling is larger in sea water and also under high pressure; such conditions exist
in deep sea-water structures. At a pressure of 10 MPa (which corresponds to a depth
of 100 m), the magnitude of swelling after 3 years can be about eight times higher
than at atmospheric pressure.9.10 Swelling which entails movement of sea water into
concrete has implications for the ingress of chlorides into concrete (see p. 569).
Drying shrinkage

461
Withdrawal of water from concrete stored in unsaturated air causes drying shrinkage.
A part of this movement is irreversible and should be distinguished from the
reversible moisture movement caused by alternating storage under wet and dry
conditions.
Mechanism of shrinkage
The change in the volume of drying concrete is not equal to the volume of water
removed. The loss of free water, which takes place first, causes little or no shrinkage.
As drying continues, adsorbed water is removed and the change in the volume of
unrestrained hydrated cement paste at that stage is equal approximately to the loss of a
water layer one molecule thick from the surface of all gel particles. Since the
‘thickness’ of a water molecule is about 1 per cent of the gel particle size, a linear
change in dimensions of cement paste on complete drying would be expected9.18 to be
of the order of 10 000 × 10–6; values up to 4000 × 10–6 have actually been observed.9.19
The influence of the gel particle size on drying is shown by the low shrinkage of
the much more coarse-grained natural building stones (even when highly porous) and
by the high shrinkage of fine grained shale.9.18 Also, high-pressure steam-cured cement
paste, which is microcrystalline and has a low specific surface, shrinks 5 to 10
times,9.14 and sometimes even 17 times,9.20 less than a similar paste cured normally.
It is possible also that shrinkage, or a part of it, is related to the removal of
intracrystalline water. Calcium silicate hydrate has been shown to undergo a change
in lattice spacing from 1.4 to 0.9 nm on drying;9.21 hydrated C3A and calcium
sulfoaluminate show similar behaviour.9.22 It is thus not certain whether the moisture
movement associated with shrinkage is inter- or intracrystalline. But, because pastes
made with both Portland and high-alumina cements, and also with pure ground
calcium monoaluminate, exhibit essentially similar shrinkage, the fundamental cause
of shrinkage must be sought in the physical structure of the gel rather than in its
chemical and mineralogical character.9.22
The relation between the mass of water lost and shrinkage is shown in Fig. 9.8. For
neat cement pastes, the two quantities are proportional to one another as no capillary
water is present and only adsorbed water is removed. However, mixes to which
pulverized silica has been added and which, for workability reasons, require a higher
water/cement ratio, contain capillary pores even when completely hydrated. Emptying
of the capillaries causes a loss of water without shrinkage but, once the capillary
water has been lost, the removal of adsorbed water takes place and causes shrinkage
in the same manner as in a neat cement paste. Thus, the final slope of all the curves
of Fig. 9.8 is the same. With concretes which contain some water in aggregate pores
and in large (accidental) cavities, an even greater variation in the shape of the curves
of water loss vs. shrinkage is found.

462
 Fig. 9.8. Relation between shrinkage and loss of water from specimens of
cement–pulverized silica pastes cured for 7 days at 21 °C (70 °F) and then
dried9.18
In concrete specimens, the loss of water with time depends on the size of specimen.
A generalized pattern of loss of water with distance from the drying surfaces was
developed by Mensi et al.9.75 on the assumption that the rate of diffusion of vapour is
proportional to the square root of time elapsed. They suggested that what occurs in a
cylinder of diameter D1 at time t1 will occur in a geometrically similar cylinder of
diameter kD1 at time k2t1. In full-size concrete members, the situation is less simple
because of the presence of edges9.55 (see Fig. 9.9). Data on the time required for
concrete, drying from one surface only, to lose 80 per cent of the evaporable water are
given in Table 9.2.

463
 Fig. 9.9. Water loss in prisms of various sizes (relative humidity of air: 55 per
cent)9.55

Table 9.2. Indicative Periods of Drying of Concrete*9.56

*
For the purpose of this table, drying is defined as loss of 80 per cent of
evaporable water.

464
 In translating the data on the loss of water into shrinkage, there is a further
complication in that, whereas in small laboratory specimens surface cracking is
minimal and potential shrinkage is achieved, in full-size structural members, surface
cracking affects the effective shrinkage and causes a redistribution of internal stresses.
Cracking possibly also increases the rate of loss of water. The topic of the influence of
the size of a concrete member upon shrinkage is considered on p. 439.
Factors influencing shrinkage
As far as shrinkage of the hydrated cement paste itself is concerned, shrinkage is
larger the higher the water/cement ratio because the latter determines the amount of
evaporable water in the cement paste and the rate at which water can move towards
the surface of the specimen. Brooks9.77 demonstrated that shrinkage of hydrated cement
paste is directly proportional to the water/cement ratio between the values of about 0.2
and 0.6. At higher water/cement ratios, the additional water is removed upon drying
without resulting in shrinkage9.77 (cf. Fig. 9.8).
Let us now turn to mortar and concrete. Table 9.3 gives typical values of drying
shrinkage of mortar and concrete specimens, 127 mm (5 in.) square in cross-section,
stored at a temperature of 21 °C (70 °F) and a relative humidity of 50 per cent for six
months. These values are no more than a guide because shrinkage is influenced by
many factors.

Table 9.3. Typical Values of Shrinkage of Mortar and Concrete Specimens, 5 in.
(127 mm) Square in Cross-section, Stored at a Relative Humidity of 50 per cent
and 21 °C (70 °F)9.19

The most important influence is exerted by aggregate, which restrains the amount
of shrinkage that can actually be realized. The ratio of shrinkage of concrete, Sc, to
shrinkage of neat cement paste, Sp, depends on the aggregate content in the concrete, a,
and is9.23

Sc = Sp(l – a)n.

The experimental values of n vary between 1.2 and 1.7,9.14 some variation arising from
the relief of stress in the cement paste by creep.9.35 Figure 9.10 shows typical results
and yields a value of n = 1.7.

465
Fig. 9.10. Influence of the aggregate content in concrete (by volume) on the ratio
of the shrinkage of concrete to the shrinkage of neat cement paste9.23
The validity of estimating the shrinkage of concrete from the shrinkage of neat
cement paste, having the same water/cement ratio and the same degree of hydration,
by taking into consideration the aggregate content and the modulus of elasticity of the
aggregate, was confirmed by Hansen and Almudaiheem.9.72
The size and grading of aggregate per se do not influence the magnitude of
shrinkage, but a larger aggregate permits the use of a leaner mix and, hence, results in
a lower shrinkage. If changing the maximum aggregate size from 6.3 to 152 mm ( in.
to 6 in.) means that the aggregate content can rise from 60 to 80 per cent of the total
volume of concrete, then, as shown in Fig. 9.10, a threefold decrease in shrinkage will
result.
Similarly, for a given strength, concrete of low workability contains more
aggregate than a mix of high workability made with aggregate of the same size and, as
a consequence, the former mix exhibits lower shrinkage.9.18 For instance, increasing the
aggregate content of concrete from 71 to 74 per cent (at the same water/cement ratio)
will reduce shrinkage by about 20 per cent (Fig. 9.10).
The twin influences of water/cement ratio and aggregate content (Table
9.3 and Fig. 9.10) can be combined in one graph; this is done in Fig. 9.11, but it must
be remembered that the shrinkage values given are no more than typical for drying in
a temperate climate. In practical terms, at a constant water/cement ratio, shrinkage
increases with an increase in the cement content because this results in a larger
volume of hydrated cement paste which is liable to shrinkage. However, at a given
workability, which approximately means a constant water content, shrinkage is
unaffected by an increase in the cement content, or may even decrease, because the

466
water/cement ratio is reduced and the concrete is, therefore, better able to resist
shrinkage. The overall pattern of these influences on shrinkage9.76 is shown in Fig. 9.12.

Fig. 9.11. Influence of water/cement ratio and aggregate content on shrinkage9.48

Fig. 9.12. The pattern of shrinkage as a function of cement content, water

content, and water/cement ratio; concrete moist-cured for 28 days, thereafter
dried for 450 days9.76
The water content of concrete affects shrinkage in so far as it reduces the volume
of restraining aggregate. Thus, in general, the water content of a mix would indicate
the order of shrinkage to be expected, following the general pattern of Fig. 9.13, but

467
the water content per se is not a primary factor. In consequence, mixes having the
same water content, but widely varying composition, may exhibit different values of
shrinkage.9.82

Fig. 9.13. Relation between the water content of fresh concrete and drying
shrinkage9.25
Let us now return to the restraining effect of the aggregate on shrinkage. The
elastic properties of aggregate determine the degree of restraint offered; for example,
steel aggregate leads to shrinkage one-third less, and expanded shale to one-third
more, than ordinary aggregate.9.6 This influence of aggregate was confirmed by
Reichard9.49 who found a correlation between shrinkage and the modulus of elasticity
of concrete, which depends on the compressibility of the aggregate used (Fig. 9.14).

468
The presence of clay in aggregate lowers its restraining effect on shrinkage and,
because clay itself is subject to shrinkage, clay coatings on aggregate can increase
shrinkage by up to 70 per cent.9.18

Fig. 9.14. Relation between drying shrinkage after 2 years and secant modulus of
elasticity of concrete (at a stress/strength ratio of 0.4) at 28 days9.49
Even within the range of ordinary aggregates, there is a considerable variation in
shrinkage of the resulting concrete (Fig. 9.15). The usual natural aggregate itself is not
normally subject to shrinkage, but there exist rocks which shrink on drying up to 900
× 10–6; this is about the same magnitude as shrinkage of concrete made with non-
shrinking aggregate. Shrinking aggregates are widespread in parts of Scotland but
they exist also elsewhere. They are mainly some dolerites and basalts, and also some
sedimentary rocks such as greywacke and mudstone. On the other hand, granite,
limestone, and quartzite have been consistently found to be non-shrinking.

469
 Fig. 9.15. Shrinkage of concretes of fixed mix proportions but made with
different aggregates, and stored in air at 21 °C (70 °F) and a relative humidity of
50 per cent.9.24 Time reckoned since end of wet curing at the age of 28 days
Concrete made with shrinking aggregate, and which therefore exhibits high
shrinkage, may lead to serviceability problems in structures due to excessive
deflection or warping (curling); if high shrinkage leads to cracking, durability of the
structure may also be impaired. For these reasons, it is useful to determine the
shrinkage of any suspect aggregate; a test method is prescribed by BS 812-120 : 1983
in which the shrinkage of concrete of fixed proportions and containing the given
aggregate is determined by drying at 105 °C. The test is not intended to be used
routinely. In this connection, it is useful to note that shrinking rocks usually have also
high absorption, and this can be treated as a warning sign that the aggregate should be
carefully investigated for its shrinkage properties. One possible way of dealing with
such aggregates is to blend high- and low-shrinkage aggregates.
Lightweight aggregate usually leads to higher shrinkage, largely because the
aggregate, having a lower modulus of elasticity, offers less restraint to the potential
shrinkage of the cement paste. Those lightweight aggregates that have a large
proportion of fine material smaller than 75 μm (No. 200) sieve have still higher
shrinkage, as the fines lead to a larger void content.
The properties of cement have little influence on the shrinkage of concrete, and
Swayze9.26 has shown that a higher shrinkage of neat cement paste does not necessarily
mean a higher shrinkage of concrete made with a given cement. Fineness of cement is
a factor only in so far as particles coarser than, say, 75 μm (No. 200) sieve, which
hydrate comparatively little, have a restraining effect similar to aggregate. Otherwise,
contrary to some earlier suggestions, finer cement does not increase shrinkage of
concrete made with normal9.26,9.41 or lightweight9.106 aggregate, although shrinkage of neat

470
cement paste is increased.9.40 The chemical composition of cement is now believed not
to affect shrinkage except that cements deficient in gypsum exhibit a greatly increased
shrinkage9.27 because the initial framework established in setting determines the
subsequent structure of the hydrated paste9.22 and thus influences also the gel/space
ratio, strength, and creep. An optimum gypsum content from the standpoint of
retardation of cement is somewhat lower than that leading to least shrinkage.9.28 For
any given cement, the range of gypsum contents which is satisfactory for shrinkage is
narrower than that for setting time.
Shrinkage of concrete made with high-alumina cement is of the same magnitude as
when Portland cement is used, but it takes place much more rapidly.9.19
Including either fly ash or ground granulated blastfurnace slag in the mix increases
shrinkage. Specifically, at a constant water/cement ratio, a higher proportion of fly
ash or slag in the blended cement leads to higher shrinkage by some 20 per cent with
the former material, and by up to 60 per cent at very high contents of slag.9.71 Silica
fume increases the long-term shrinkage.9.81
Water-reducing admixtures per se probably cause a small increase in shrinkage.
Their main effect is indirect in that the use of an admixture may result in a change in
the water content or in the cement content of the mix, or in both, and it is the
combined action of those changes that influences shrinkage.9.71 Superplasticizers have
been found9.71 to increase shrinkage by some 10 to 20 per cent. However, the changes
in the observed shrinkage are too small to be accepted as reliable and generally valid.
From the preceding statements, it can be expected that shrinkage of very-high-
strength concrete, which contains a superplasticizer, is simply the outcome of the
relevant and opposing factors: a very low water/cement ratio and concomitant high
self-desiccation, which lead to low shrinkage, and a high cement content, which leads
to high shrinkage. Thus, the usual approach to estimating shrinkage applies also to
very-high-strength concrete. However, the more rigid structure of such concrete
restrains the magnitude of the effective shrinkage.
Entrainment of air has been found to have no effect on shrinkage.9.29 Added calcium
chloride increases shrinkage, generally between 10 and 50 per cent,9.30 probably
because a finer gel is produced and possibly because of greater carbonation of the
more mature specimens with calcium chloride.9.50
Influence of curing and storage conditions
Shrinkage takes place over long periods: some movement has been observed even
after 28 years9.24 (Fig. 9.16), but a part of the long-term shrinkage is likely to be due to
carbonation. Figure 9.16 (in which time is plotted on a logarithmic scale) shows that
the rate of shrinkage decreases rapidly with time.

471
 Fig. 9.16. Range of shrinkage–time curves for different concretes stored at
relative humidities of 50 and 70 per cent9.24
Prolonged moist curing delays the advent of shrinkage, but the effect of curing on
the magnitude of shrinkage is small, though rather complex. As far as neat cement
paste is concerned, the greater the quantity of hydrated cement the smaller is the
volume of unhydrated cement particles which restrain the shrinkage: thus prolonged
curing could be expected to lead to greater shrinkage,9.18 but the hydrated cement paste
contains less water and becomes stronger with age and is able to attain a larger
fraction of its shrinkage tendency without cracking. However, in concrete, if cracking
takes place, e.g. around aggregate particles, the overall shrinkage, measured on a
concrete specimen, apparently decreases. Well-cured concrete shrinks more
rapidly9.40 and, therefore, the relief of shrinkage stresses by creep is smaller; also, the
concrete, being stronger, has an inherent low creep capacity. These factors may
outweigh the higher tensile strength of well-cured concrete and may lead to cracking.
In view of this, it is not surprising that contradictory results on the effects of curing on
shrinkage have been reported, but in general the length of the curing period is not an
important factor in shrinkage.
The magnitude of shrinkage is largely independent of the rate of drying except that
transferring concrete directly from water to a very low humidity can lead to fracture.
Rapid drying out does not allow a relief of stress by creep and may lead to more
pronounced cracking. However, neither wind nor forced convection have any effect
on the rate of drying of hardened concrete (except during very early stages) because
the moisture conductivity of concrete is so low that only a very small rate of
evaporation is possible: the rate cannot be increased by movement of air.9.51 This has
been confirmed experimentally.9.52 (See p. 319 for evaporation from fresh concrete.)
The relative humidity of the medium surrounding the concrete greatly affects the
magnitude of shrinkage, as shown for instance in Fig. 9.17. The same figure illustrates
also the greater absolute magnitude of shrinkage compared with swelling in water:

472
swelling is about six times smaller than shrinkage in air of relative humidity of 70 per
cent or eight times smaller than shrinkage in air at 50 per cent.

Fig. 9.17. Relation between shrinkage and time for concretes stored at different
relative humidities.9.24 Time reckoned since end of wet curing at the age of 28 days
We see thus that concrete placed in ‘dry’ (unsaturated) air shrinks, but it swells in
water or air with a relative humidity of 100 per cent. This would indicate that the
vapour pressure within the cement paste is always less than the saturated vapour
pressure, and it is logical to expect that there is an intermediate humidity at which the
paste would be in hygral equilibrium. In fact, Lorman9.31 found this humidity to be 94
per cent, but in practice equilibrium is possible only in small and practically
unrestrained specimens.
When it is desired to estimate shrinkage at a given relative humidity on the basis of
a known value of shrinkage at some other relative humidity, the relation of ACI
209R-929.80 can be used. This is shown in Fig. 9.18, which includes also the relation
proposed by Hansen and Almudaiheem.9.72 The latter indicates a relative shrinkage
lower than that given by ACI 209R-92 at relative humidities above 50 per cent.
Hansen and Almudaiheem9.72 also give values of relative shrinkage in the range of
relative humidity of 11 to 40 per cent, for which ACI 209R-92 does not give any
specific values.

473
Fig. 9.18. Relative value of shrinkage as a function of ambient relative humidity
according to ACI 209R-929.80 and Hansen and Almudaiheem9.72
Prediction of shrinkage
According to ACI 209R-92,9.80 the development of shrinkage with time follows the
equation

where st = shrinkage after t days since the end of 7-day moist curing
sult = ultimate shrinkage, and
t = time in days since the end of moist curing.
Prediction of the development of shrinkage by the above equation is subject to
considerable variability, but the equation can be used to estimate ultimate shrinkage
of a wide range of moist-cured concretes. It can be seen that one-half of the ultimate
shrinkage is expected to occur after 35 days’ drying. For steam-cured concrete, the
value of 35 in the denominator is replaced by 55, and time t is reckoned from the end
of steam curing at 1 to 3 days.
ACI 209R-929.80 gives a general expression for the prediction of shrinkage by
modifying a standard value by a number of coefficients which allow for various
factors. The error involved in such an approach must be expected to be large.
Various expressions for shrinkage are discussed by Neville et al.9.84 These
expressions can be used to estimate long-term shrinkage from short-term tests on the

474
actual concrete. Such tests are necessary for a reasonably accurate prediction of
shrinkage.
A method for determination of short-term shrinkage is prescribed in BS 1881-5 :
1984: the specimens are dried for a specified period under prescribed conditions of
temperature and humidity. The shrinkage occurring under these conditions is about
the same as that after a long exposure to air with a relative humidity of approximately
65 per cent,9.19 and is therefore in excess of the shrinkage encountered outdoors in the
British Isles. The magnitude of shrinkage can be determined using a measuring frame
fitted with a micrometer gauge or a dial gauge reading to 10–5 strain, or by means of an
extensometer or strain gauges. The American test method is prescribed by ASTM C
157-93; the air movement past the test specimens is carefully controlled and the
relative humidity is maintained at 50 per cent. The ISO method is described in BS
ISO 1920 : 2009.
Differential shrinkage
It was mentioned earlier that the potential shrinkage of neat cement paste is restrained
by the aggregate. In addition, some restraint arises also from non-uniform shrinkage
within the concrete member itself. Moisture loss takes place only at the surface so that
a moisture gradient is established in the concrete specimen, which is thus subjected to
differential shrinkage. The potential shrinkage is compensated by the strains due to
internal stresses, tensile near the surface and compressive in the core. When drying
takes place in an unsymmetrical manner, warping (curling) can result.
It may be useful to point out that the values of shrinkage generally quoted are those
of free shrinkage, or potential shrinkage, that is, contraction unrestrained either
internally or by external constraints on a structural member. In considering the effect
of the constraining forces on the actual shrinkage, it is important to realize that the
induced stresses are modified by relaxation, which may prevent the development of
cracking, as discussed on p. 442. Because relaxation occurs only slowly, it may
prevent cracking when shrinkage develops slowly; however, the same magnitude of
shrinkage occurring rapidly may well induce cracking. It is shrinkage cracking that is
of paramount interest.
The progress of shrinkage extends gradually from the drying surface into the
interior of the concrete but does so only extremely slowly. Desiccation was observed
to reach the depth of 75 mm (3 in.) in one month but only 600 mm (2 ft) after 10
years.9.14 Data9.55 of L’Hermite are shown in Fig. 9.19; initial swelling in the interior can
be seen. Ross9.32 found the difference between shrinkage in a mortar slab at the surface
and at a depth of 150 mm (6 in.) to be 470 × 10–6 after 200 days. If the modulus of
elasticity of mortar is 21 GPa (3 × 106 psi) the differential shrinkage would induce a
stress of 10 MPa (1400 psi); because the stress arises gradually it is relieved by creep
but, even so, surface cracking may result.

475
Fig. 9.19. Progress of shrinkage with time as a function of distance from drying
surface (no drying possible in other directions). (Shrinkage values corrected for
temperature differences)9.55
Because drying takes place at the surface of concrete, the magnitude of shrinkage
varies considerably with the size and shape of the specimen, being a function of the
surface/volume ratio.9.32 A part of the size effect may also be due to the pronounced
carbonation shrinkage of small specimens (see p. 444). Thus, for practical purposes,
shrinkage cannot be considered as purely an inherent property of concrete without
reference to the size of the concrete member.
Many investigations have, in fact, indicated an influence of the size of the
specimen on shrinkage. The observed shrinkage decreases with an increase in the size
of the specimen but, above some value, the size effect is small initially, although
pronounced later (Fig. 9.20). The shape of the specimen also appears to enter the
picture but, as a first approximation, shrinkage can be expressed as a function of the

476
volume/surface ratio of the specimen. There appears to be a linear relation between
this ratio and the logarithm of shrinkage9.53 (Fig. 9.21). Furthermore, the ratio is
linearly related to the logarithm of time required for half the shrinkage to be achieved.
The latter relation applies to concretes made with different aggregates, so that,
whereas the magnitude of shrinkage is affected by the type of aggregate used, the rate
at which the final value of shrinkage is reached is not influenced.9.53 It has been
argued9.16,9.83 that theoretically the ultimate shrinkage is independent of the size of the
concrete element but, for realistic periods, it must be accepted that shrinkage is
smaller in larger elements.

Fig. 9.20. Relation between axial shrinkage and width of concrete prisms of
square cross-section and length/width ratio of 4 (drying allowed at all surfaces)9.55

477
 Fig. 9.21. Relation between ultimate shrinkage and volume/surface ratio9.53
The effect of shape is secondary. I-shaped specimens exhibit less shrinkage than
cylindrical ones of the same volume/surface ratio, the difference being 14 per cent on
the average.9.53 The difference, which can be explained in terms of variation in the
mean distance that the water has to travel to the surface, is thus not significant for
design purposes.
Shrinkage-induced cracking
As mentioned in connection with differential shrinkage, the importance of shrinkage
in structures is largely related to cracking. Strictly speaking, we are concerned with
the cracking tendency because the advent or absence of cracking depends not only on
the potential contraction but also on the extensibility of concrete, its strength, and its
degree of restraint to the deformation that may lead to cracking.9.54 Restraint in the
form of reinforcing bars or a gradient of stress increases extensibility of concrete in
that it allows it to develop strain well beyond that corresponding to maximum stress.
A high extensibility of concrete is generally desirable because it permits concrete to
withstand greater volume changes.
The schematic pattern of crack development when stress is relieved by creep is
shown in Fig. 9.22. Cracking can be avoided only if the stress induced by the free
shrinkage strain, reduced by creep, is at all times smaller than the tensile strength of
the concrete. Thus, time has a two-fold effect: the strength increases, thereby reducing
the danger of cracking but, on the other hand, the modulus of elasticity also increases
so that the stress induced by a given shrinkage becomes larger. Furthermore, the creep
relief decreases with age so that the cracking tendency becomes greater. A minor
practical point is that, if the cracks due to restrained shrinkage form at an early stage,
and moisture subsequently has access to the crack, many of the cracks will become
closed by autogenous healing.

478
 Fig. 9.22. Schematic pattern of crack development when tensile stress due to
restrained shrinkage is relieved by creep
One of the most important factors in cracking is the water/cement ratio of the mix
because its increase tends to increase shrinkage and, at the same time, to reduce the
strength of the concrete. An increase in the cement content also increases shrinkage
and, therefore, the cracking tendency, but the effect on strength is positive. This
applies to drying shrinkage. Carbonation, although it produces shrinkage, reduces
subsequent moisture movement, and therefore is advantageous from the standpoint of
cracking tendency. On the other hand, the presence of clay in aggregate leads both to
higher shrinkage and to greater cracking.
The use of admixtures may influence the cracking tendency through an interplay of
effects on hardening, shrinkage, and creep. Specifically, retarders may allow more
shrinkage to be accommodated in the form of plastic shrinkage (see p. 424) and also
probably increase the extensibility of concrete, and therefore reduce cracking. On the
other hand, if concrete has attained rigidity too rapidly, it cannot accommodate the
would-be plastic shrinkage and, having a low strength, cracks.
The temperature at the time of placing determines the dimensions of concrete at
the moment when it ceases to deform plastically (i.e. without loss of continuity). A
subsequent drop in temperature will produce potential contraction. Thus, placing
concrete in hot weather means a high cracking tendency. Steep temperature or
moisture gradients produce severe internal restraints and thus represent a high
cracking tendency. Likewise, restraint by the base of a member, or by other members,
may lead to cracking.
These are some of the factors to be considered. Actual cracking and failure depend
on the combination of factors, and indeed it is rarely that a single adverse factor is
responsible for cracking of concrete. Early age thermal cracking is considered by
CIRIA.9.160
There exists no standard test to assess cracking due to restrained shrinkage, but the
use of a ring-shaped concrete specimen restrained by an internal steel ring can be
informative with respect to the comparative resistance of different concretes to
cracking.9.78,9.79 Cracking of concrete due to various causes is considered in Chapter 10.

479
Moisture movement
If concrete which has been allowed to dry in air of a given relative humidity is
subsequently placed in water (or at a higher humidity) it will swell. Not all initial
drying shrinkage is, however, recovered, even after prolonged storage in water. For
the usual range of concretes, the irreversible part of shrinkage represents between 0.3
and 0.6 of the drying shrinkage,9.14 the lower value being more common.9.25 The
absence of fully reversible behaviour is probably due to the introduction of additional
bonds within the gel during the period of drying, when closer contact between the gel
particles is established. If the cement paste has hydrated to a considerable degree
before drying, it will be less affected by the closer configuration of the gel when dry;
in fact, neat cement paste, water-cured for six months and then dried, was found to
have no residual shrinkage on rewetting.9.33 On the other hand, if drying is
accompanied by carbonation, the cement paste becomes insensitive to moisture
movement so that the residual shrinkage is increased.9.14
The influence on moisture movement of curing before drying and of carbonation
during drying may explain why there is no simple relation between the magnitude of
moisture movement and shrinkage.
Figure 9.23 shows the moisture movement, expressed as linear strain, of cement
paste subjected to alternating storage in water and in air at a relative humidity of 50
per cent.9.33 The magnitude of the moisture movement varies with the range of
humidity and composition of concrete (Table 9.4). Lightweight concrete has a higher
moisture movement than concrete made with ordinary aggregate.

Fig. 9.23. Moisture movement of a 1:1 cement : pulverized basalt mix stored
alternately in water and in air at 50 per cent relative humidity; cycle period 28
days9.33

Table 9.4. Typical Values of Moisture Movement of Mortar and Concrete Dried
at 50 °C (122 °F) and Immersed in Water9.19

480
 For a given concrete, there is a gradual reduction in the moisture movement during
succeeding cycles, probably due to the creation of additional bonds within the gel.9.22 If
the water storage periods are of sufficient duration, the continued hydration of cement
results in some additional swelling so that there is a net increase in dimensions
superimposed on the reversible movement due to drying and wetting.9.19 (In Fig.
9.23 this would be shown by a slight rise in the upper dotted line.)
Carbonation shrinkage
In addition to shrinkage upon drying, the surface zone of concrete undergoes
shrinkage due to carbonation, and some of the experimental data on drying shrinkage
include the effects of carbonation. Drying shrinkage and carbonation shrinkage are,
however, quite distinct in nature.
The process of carbonation is discussed in Chapter 10 and, at this stage, our
concern is limited to carbonation shrinkage. However, it should be noted that, because
carbon dioxide is fixed by the hydrated cement paste, the mass of the latter increases.
Consequently, the mass of concrete also increases. When concrete dries and
carbonates simultaneously, the increase in mass on carbonation may at some stage
give the misleading impression that the drying process has reached the stage of
constant mass, i.e. equilibrium (see Fig. 9.24). Such an interpretation of test data must
clearly be guarded against.9.58

481
 Fig. 9.24. Loss of mass of concrete due to drying and carbonation9.58
Carbonation shrinkage is probably caused by the dissolving of crystals of
Ca(OH)2 while under a compressive stress (imposed by the drying shrinkage) and
depositing of CaCO3 in spaces free from stress; the compressibility of the hydrated
cement paste is thus temporarily increased. If carbonation proceeds to the stage of
dehydration of C-S-H, this also produces carbonation shrinkage.9.104
Figure 9.25 shows the drying shrinkage of mortar specimens dried in CO2-free air
at different relative humidities, and also the shrinkage after subsequent carbonation.
Carbonation increases the shrinkage at intermediate humidities, but not at 100 per
cent or 25 per cent. In the latter case, there is insufficient water in the pores within the
cement paste for CO2 to form carbonic acid. On the other hand, when the pores are
full of water, the diffusion of CO2 into the paste is very slow; it is also possible that
the diffusion of calcium ions from the paste leads to precipitation of CaCO3 with a
consequent clogging of surface pores.9.37

482
 Fig. 9.25. Drying shrinkage and carbonation shrinkage of mortar at different
relative humidities9.37
The sequence of drying and carbonation greatly affects the total magnitude of
shrinkage. Simultaneous drying and carbonation produces lower total shrinkage than
when drying is followed by carbonation (Fig. 9.26) because, in the former case, a
large part of the carbonation occurs at relative humidities above 50 per cent: under
such conditions carbonation shrinkage is reduced (Fig. 9.25). Carbonation shrinkage
of high-pressure steam-cured concrete is very small.

483
 Fig. 9.26. Influence of the sequence of drying and carbonation of mortar on
shrinkage9.37
When concrete is subjected to alternating wetting and drying in air containing CO2,
shrinkage due to carbonation (during the drying cycle) becomes progressively more
apparent. The total shrinkage at any stage is greater than if drying took place in CO2-
free air,9.37 so that carbonation increases the magnitude of irreversible shrinkage and
may contribute to crazing of exposed concrete. Crazing is a form of shallow cracking
induced by the restrained shrinkage of the surface zone against the non-shrinking
interior of the concrete.
However, carbonation of concrete prior to exposure to alternating wetting and
drying reduces the moisture movement, sometimes by nearly a half.9.38 A practical
application of this is to pre-carbonate precast products immediately after demoulding
by exposing them to flue gases. Concrete with a small moisture movement is then
obtained, but the humidity conditions during carbonation have to be carefully
controlled. Various techniques of carbonation of concrete products are described in
ACI 517.2R-87.9.96

Shrinkage compensation by the use of expansive cements*

*
This section was substantially published in ref. 9.105.
The discussion of drying shrinkage earlier in this chapter should have made it clear
that shrinkage is probably one of the least desirable properties of concrete. When

484
shrinkage is restrained, it may lead to shrinkage cracking, which mars the appearance
of concrete and makes it more vulnerable to attack by external agents, thus adversely
affecting durability. But even unrestrained shrinkage is harmful: adjacent concrete
elements shrink away from one another, thus opening ‘external cracks’. Shrinkage is
also responsible for a part of the loss of the initial stress in the tendons in prestressed
concrete.
It is not surprising, therefore, that many attempts have been made to develop a
cement which, on hydration, would counteract the deformation induced by shrinkage.
In special cases, even a net expansion of concrete on hardening may be advantageous.
Concrete containing such an expansive cement expands in the first few days of its life,
and a form of prestress is obtained by restraining this expansion with steel
reinforcement: steel is put in tension and concrete in compression. Restraint by
external means is also possible. Such concrete is known as shrinkage-compensating
concrete.
It is also possible to use expansive cement in order to produce self-stressing
concrete, in which the restrained expansion, remaining after most of the shrinkage has
occurred, is high enough to induce a significant compressive stress in concrete9.3 (up to
about 7 MPa (1000 psi)).
Expansive cement, although considerably more expensive than Portland cement, is
valuable in concrete structures in which a reduction in cracking is of importance, for
instance, bridge decks, pavement slabs, and liquid storage tanks.
It is worth making it clear that the use of expansive cement does not prevent the
development of shrinkage. What happens is that the restrained early expansion
balances approximately the subsequent normal shrinkage; this is shown in Fig. 9.27.
Usually, a small residual expansion is aimed at because, as long as some compressive
stress in concrete is retained, shrinkage cracking will not develop.

485
 Fig. 9.27. Diagrammatic representation of length changes of shrinkage-
compensating and Portland cement concretes (based on ref. 9.91)
Types of expansive cements
Early development of expansive cements took place in Russia and in France, where
Lossier9.2 used a mixture of Portland cement, an expanding agent, and a stabilizer. The
expanding agent was obtained by burning a mixture of gypsum, bauxite, and chalk,
which form calcium sulfate and calcium aluminate (mainly C5A3). In the presence of
water, these compounds react to form calcium sulfoaluminate hydrate (ettringite),
with an accompanying expansion of the cement paste. The stabilizer, which is
blastfurnace slag, slowly takes up the excess calcium sulfate and brings expansion to
an end.
Nowadays, three main types of expansive cement are produced, but only one, Type
K, is commercially available in the United States. ASTM C 845-04 classifies
expansive cements, collectively referred to as Type E-1, according to the expansive
agent used with Portland cement and calcium sulfate. In each case, the agent is a
source of reactive aluminate which combines with the sulfates in the Portland cement
to form expansive ettringite; for instance, in Type K cement, the reaction is

4CaO . 3A12O3 . SO3 + 8[CaO . SO3 . 2H2O] + 6[CaO . H2O] + 74H2O →

3[3CaO . A12O3 . 3CaSO4 . 32H2O].

The resulting compound is known as ettringite.

Calcium sulfate reacts rapidly with 4CaO . 3A12O3 . SO3 because it exists in a
separate form,9.85 unlike C3A which is part of the Portland cement clinker.
Whereas the formation of ettringite in mature concrete is harmful (see p. 511), a
controlled formation of ettringite in the early days after placing of concrete is used to
achieve a shrinkage-compensating effect.
The three types of expansive cement recognized by ACI 223R-939.91 and by ASTM
C 845-04, are:
Type K which contains 4CaO . 3A12O3 . SO3 and uncombined CaO,
Type M which contains calcium aluminates CA and C12A7, and
Type S which contains C3A in excess of the amount normally present in
Portland cement.
In addition, in Japan, there is produced an expansive cement which uses specially
processed calcium oxide9.8 to produce free-lime expansion. This cement is called Type
O.
Type K cement is produced by integral burning of the components or by
intergrinding. It is also possible, as is done in Japan,9.8 to add the expansive component
at the concrete batching plant.
Special expansive cements, containing high-alumina cement, for particular
purposes where an extremely high expansion is required, can also be produced.9.92
Shrinkage-compensating concrete
The expansion of cement paste resulting from the formation of ettringite begins as
soon as water has been added to the mix, but only restrained expansion is beneficial,
and no restraint is offered while concrete is in the plastic state or while it has

486
negligible strength. For this reason, prolonged mixing9.86 and delay before placing of
concrete containing expansive cement should be avoided.
On the other hand, delayed expansion in concrete in service may prove disruptive,
as is the case with external sulfate attack (see p. 511). It is, therefore, important that
ettringite formation ceases after several days, and this happens when either SO3 or
A12O3 has become exhausted.
ASTM C 845-04 prescribes a maximum 7-day expansion of mortar of between 400
× 10–6 and 1000 × 10–6; the 28-day expansion must not exceed the 7-day expansion by
more than 15 per cent. The latter value is a check on delayed expansion.
Because the formation of ettringite requires a large amount of water, wet curing of
concrete made with expansive cement is necessary for full benefits of the use of such
cement to be reaped.9.87
Information on the use of expansive cements so as to obtain shrinkage-
compensating concrete is given in ACI 223R-939.91 but some features of this type of
concrete merit mention here. Its water requirement is about 15 per cent higher than
when Portland cement only is used. However, as some of this additional water
becomes combined very early, the strength of concrete is little affected.9.91 Another
way of representing the situation is to say that, at the same water/cement ratio,
concrete made with Type K expansive cement has a 28-day compressive strength
some 25 per cent higher than concrete made with Portland cement only.9.4,9.85
At a given water content, the workability of expansive cement concrete is lower
and the slump loss is greater.9.86
The usual admixtures can be used in shrinkage-compensating concrete but trial
mixes are necessary because some admixtures, especially air-entraining ones, may not
be compatible with certain expansive cements.9.55,9.86
Because expansive cement has a large content of calcium sulfate, which is softer
than Portland cement clinker, the cement has a high specific surface, typically 430
kg/m2. Excessive fineness, by promoting rapid hydration, may lead to premature
expansion,9.91 which is ineffective because very young concrete is unable to offer
restraint. The expansion is greater the higher the cement content of the concrete and
the higher the modulus of elasticity of the aggregate9.3 because the aggregate offers
restraint to the expansion of the cement paste. ASTM 878-09 prescribes a test method
for restrained expansion of shrinkage-compensating concrete. This test can be used to
study the effects of various factors on expansion.
Silica fume can be incorporated into shrinkage-compensating concrete in order to
control excessive expansion.9.90 Tests on Type K cement paste9.89 have shown that silica
fume in the mix accelerates expansion but the expansion stops before CaO . 3A12O3 .
SO3 has been used up, probably due to a lowering of the pH. The absence of long-term
expansion is desirable and shortening the wet-curing period to 4 days is convenient.
If, following the expansive reactions, the cement is undersulfated, the concrete is
vulnerable to sulfate attack (see p. 509); this may be the case with Type M and Type S
cements.9.4

Creep of concrete*
*
For fuller treatment of this topic, see A. M. Neville, W. Dilger and J. J.
Brooks, Creep of Plain and Structural Concrete (London, Construction Press,
Longman Group, 1983).

487
We have seen that the relation between stress and strain for concrete is a function of
time: the gradual increase in strain with time under load is due to creep. Creep can
thus be defined as the increase in strain under a sustained stress (Fig. 9.28) and,
because this increase can be several times as large as the strain on loading, creep is of
considerable importance in structures.

488
489
Fig. 9.28. Time-dependent deformations in concrete subjected to a sustained load
Creep may also be viewed from another standpoint: if the restraint is such that a
stressed concrete specimen is subjected to a constant strain, creep will manifest itself
as a progressive decrease in stress with time.9.107 This form of relaxation is shown
in Fig. 9.29.

Fig. 9.29. Relaxation9.107 of stress under a constant strain of 360 × 10–6

Under normal conditions of loading, the instantaneous strain recorded depends on
the speed of application of the load and includes thus not only the elastic strain but
also some creep. It is difficult to differentiate accurately between the immediate
elastic strain and early creep, but this is not of practical importance as it is the total
strain induced by the application of load that matters. Because the modulus of
elasticity of concrete increases with age, the elastic deformation gradually decreases
and, strictly speaking, creep should be taken as strain in excess of the elastic strain at
the time at which creep is being determined (Fig. 9.28). Often, the modulus of
elasticity is not determined at different ages, and creep is simply taken as an increase
in strain above the initial elastic strain. This alternative definition, although
theoretically less correct, does not introduce a serious error and is often more
convenient to use except in rigorous analysis.
So far, we have considered the creep of concrete stored under such conditions that
no shrinkage or swelling takes place. If a specimen is drying while under load, it is
usually assumed that creep and shrinkage are additive; creep is thus calculated as the
difference between the total time-deformation of the loaded specimen and the
shrinkage of a similar unloaded specimen stored under the same conditions through
the same period (Fig. 9.28). This is a convenient simplification but, as shown on

490
p. 460, shrinkage and creep are not independent phenomena to which the principle of
superposition can be applied, and in fact the effect of shrinkage on creep is to increase
the magnitude of creep. In the case of many actual structures, however, creep and
shrinkage occur simultaneously and the treatment of the two together is, from the
practical standpoint, often convenient.
For this reason, and also because the great majority of the available data on creep
were obtained on the assumption of the additive properties of creep and shrinkage, the
discussion in this chapter will, for the most part, consider creep as a deformation in
excess of shrinkage. However, where a more fundamental approach is warranted,
distinction will be made between creep of concrete under conditions of no moisture
movement to or from the ambient medium (true or basic creep) and the additional
creep caused by drying (drying creep). The terms and definitions involved are
illustrated in Fig. 9.28.
If a sustained load is removed, the strain decreases immediately by an amount
equal to the elastic strain at the given age, generally lower than the elastic strain on
loading. This instantaneous recovery is followed by a gradual decrease in strain,
called creep recovery (Fig. 9.30). The shape of the creep recovery curve is rather like
that of the creep curve, but the recovery approaches its maximum value much more
rapidly.9.108 The recovery of creep is not complete, and creep is not a simply reversible
phenomenon, so that any sustained application of load, even only over a period of a
day, results in a residual deformation. Creep recovery is of importance in predicting
deformation of concrete under a stress which varies with time.

Fig. 9.30. Creep and creep recovery of a mortar specimen, stored in air at a
relative humidity of 95 per cent, subjected to a stress of 14.8 MPa (2150 psi) and
then unloaded9.108
Factors influencing creep
In most investigations, creep has been studied empirically in order to determine how it
is affected by various properties of concrete. A difficulty in interpreting many of the
available data arises from the fact that, in proportioning concrete, it is not possible to
change one factor without altering also at least one other. For instance, the richness
and the water/cement ratio of a mix of a given workability vary at the same time.
Certain influences are, however, apparent.

491
 Some of these arise from the intrinsic properties of the mix, others from external
conditions. First of all, it should be noted that it is really the hydrated cement paste
which undergoes creep, the role of the aggregate in concrete being primarily that of
restraint; the usual normal weight aggregates are not liable to creep under the stresses
existing in concrete. The situation is thus similar to that in the case of shrinkage (see
p. 430). Creep is, therefore, a function of the volumetric content of cement paste in
concrete, but the relation is not linear. It has been shown9.109 that creep of concrete, c,
the volumetric content of aggregate, g, and the volumetric content of unhydrated
cement, u, are related by

where cp is creep of neat cement paste of the same quality as used in concrete, and

Here, μa = Poisson’s ratio of aggregate, μ = Poisson’s ratio of surrounding material

(concrete), Ea = modulus of elasticity of aggregate, and E = modulus of elasticity of
the surrounding material. This relation applies to concrete made both with normal
weight aggregate and with lightweight aggregate.9.110
Figure 9.31 illustrates the relation between creep of concrete and its aggregate
content (the volume of unhydrated cement being ignored). It may be noted that, in the
majority of the usual mixes, the variation in the aggregate content is small, but an
increase in the aggregate content by volume from 65 to 75 per cent can decrease creep
by 10 per cent.

492
 Fig. 9.31. Relation between creep c after 28 days under load and content of
aggregate g for wet-stored specimens loaded at the age of 14 days to a
stress/strength ratio of 0.509.109
The grading, maximum size, and shape of the aggregate have been suggested as
factors in creep. However, their main influence lies in the effect that they have
directly or indirectly on the aggregate content,9.109 providing that full consolidation of
concrete has been achieved in all cases.
There are certain physical properties of aggregate which influence the creep of
concrete. The modulus of elasticity of aggregate is probably the most important factor.
The higher the modulus the greater the restraint offered by the aggregate to the
potential creep of the hydrated cement paste; this is evident from the expression for α
above.
Porosity of aggregate has also been found to influence the creep of concrete but,
because aggregates with a higher porosity generally have a lower modulus of
elasticity, it is possible that porosity is not an independent factor in creep. On the
other hand, it can be visualized that the porosity of aggregate, and even more so its
absorption, play a direct role in the transfer of moisture within concrete; this transfer
may be associated with creep in that it produces conditions conducive to the
development of drying creep. This may be the explanation for the high initial creep
occurring with some lightweight aggregates batched in a dry condition.
Because of the great variation in aggregate within any mineralogical and
petrological type, it is not possible to make a general statement about the magnitude

493
of creep of concrete made with aggregates of different types. However, the data of Fig.
9.32 are of considerable importance: after 20 years’ storage at a relative humidity of
50 per cent, concrete made with sandstone aggregate exhibited creep more than twice
as great as concrete made with limestone. An even greater difference between the
creep strains of concretes made with different aggregates was found by Rüsch et
al.9.111 After 18 months under load at a relative humidity of 65 per cent, the maximum
creep was five times the minimum value, the aggregates in the increasing order of
creep being: basalt; quartz; gravel, marble and granite; and sandstone.

Fig. 9.32. Creep of concretes of fixed proportions but made with different
aggregates, loaded at the age of 28 days, and stored in air at 21 °C (70 °F) and a
relative humidity of 50 per cent9.24
There is no fundamental difference between normal and lightweight aggregates as
far as the creep properties are concerned, and the higher creep of concretes made with
lightweight aggregates reflects only the lower modulus of elasticity of that aggregate.
The rate of creep of lightweight aggregate concrete decreases with time more slowly
than in normal weight concrete. As a general rule, it can be stated that creep of
structural quality lightweight aggregate concrete is about the same as that of concrete
made with ordinary aggregate. (It is important in any comparison that the aggregate
contents do not differ widely between the lightweight and ordinary concretes.)
Furthermore, because the elastic deformation of lightweight aggregate concrete is
usually larger than in ordinary concrete, the ratio of creep to elastic deformation is
smaller for lightweight aggregate concrete.9.112
Influence of stress and strength
At this stage, it may be appropriate to consider the influence of stress on creep. There
is a direct proportionality between creep and the applied stress,9.113 with a possible

494
exception of specimens loaded at a very early age. There is no lower limit of
proportionality because concrete undergoes creep even at a very low stress. The upper
limit of proportionality is reached when severe microcracking develops in concrete;
this occurs at a stress, expressed as a fraction of strength, which is lower in a more
heterogeneous material. Thus, the limit in concrete is usually between 0.4 and 0.6, but
occasionally as low as 0.3 or as high as 0.75; the latter value applies to high strength
concrete.9.66 In mortar, the limit is in the region of 0.80 to 0.85.9.112
It appears safe to conclude that, within the range of stresses in structures in service,
the proportionality between creep and stress holds good, and creep expressions
assume this to be the case. Creep recovery is also proportional to the stress previously
applied.9.114
Above the limit of proportionality, creep increases with an increase in stress at an
increasing rate, and there exists a stress/strength ratio above which
creep produces time failure. This stress/strength ratio is in the region of 0.8 to 0.9 of
the short-term static strength. Creep increases the total strain until this reaches a
limiting value corresponding to the ultimate strain of the given concrete. This
statement implies a limiting strain concept of failure, at least in the hardened cement
paste (see p. 294).
The strength of concrete has a considerable influence on creep: within a wide
range, creep is inversely proportional to the strength of concrete at the time of
application of the load. This is indicated, for instance, in the data of Table 9.5. It is
thus possible to express creep as a linear function of the stress/strength ratio9.115 (Fig.
9.33). This proportionality has been widely confirmed. It may not be a fundamental
relationship but it is a most convenient one because, in practice, the strength of
concrete is specified and the stress under a sustained load is calculated by the designer.
For this reason, the stress/strength ratio approach is thought more practical than the
consideration of type of cement, water/cement ratio, and age. In our approach, while
we recognize the role of the water/cement ratio, we utilize the fact that, for the same
stress/strength ratio, creep is sensibly independent of the water/cement ratio. Likewise,
we ignore the age as such, its influence being mainly in increasing the strength of
concrete. It may be appropriate to note that even very old concrete undergoes creep,
as demonstrated by tests on 50-year-old concrete.9.116

Table 9.5. Ultimate Specific Creep of Concretes of Different Strengths Loaded at

the Age of 7 Days

495
Fig. 9.33. Creep of mortar specimens cured and stored continuously at different
humidities9.117
Influence of properties of cement
The type of cement affects creep in so far as it influences the strength of the concrete
at the time of application of the load. For this reason, any comparison of creep of
concretes made with different cements should take into account the influence of the
type of cement on the strength of concrete at the time of application of the load. On
this basis, both Portland cements of different types and high-alumina cement lead to
sensibly the same creep,9.123,9.124 but the rate of gain of strength has some effect as shown
below.
Fineness of cement affects the strength development at early ages and thus
influences creep. It does not seem, however, that fineness per se is a factor in creep:
contradictory results may be due to the indirect influence of gypsum. The finer the
cement the higher its gypsum requirement, so that re-grinding a cement in the
laboratory without the addition of gypsum produces an improperly retarded cement,
which exhibits high shrinkage and high creep.9.28 Extremely fine cements, with a
specific surface up to 740 kg/m2, lead to a higher early creep but to lower creep after
one or two years under load.9.41 This is probably due to the high gain of strength of the
finest cement with resultant rapid drop in the actual stress/strength ratio.9.133
The change in strength of concrete while under load is of importance in evaluating
the preceding statement that creep is not influenced by the type of cement. For the
same stress/strength ratio at the time of application of load, creep is smaller the
greater the relative increase in strength beyond the time of application of load.9.133 Thus,
creep increases in order for: low heat, ordinary, and rapid-hardening cements. There is

496
no doubt, however, that for a constant applied stress (not stress/strength ratio) at a
fixed (early) age, creep increases in order for: rapid-hardening, ordinary, and low heat
cements. These two statements bring out clearly the need for a full qualification of
information about factors in creep.
The influence on creep of the strength of concrete at the time of application of load
applies also when different cementitious materials are used.
Otherwise, quantitative generalizations about creep of concretes containing fly ash or
ground granulated blastfurnace slag are not possible because published literature
reports investigations, each of which used specific and differing test conditions. Such
data cannot be used to predict the creep of concrete at the structural design stage. All
we can say reliably is that the pattern of development of creep and of creep recovery
is not affected by the presence of Class C or Class F fly ash,9.144,9.153 ground granulated
blastfurnace slag,9.151 or silica fume, or even combinations of these materials.
There may, however, be some influences on creep of the structure of the hydrated
cement arising from inclusion of the various cementitious materials. The influence on
drying creep, where permeability and diffusivity of the hydrated cement paste are
relevant, may be different from the influence on basic creep. For instance, the use of
blastfurnace slag leads to a lower basic creep but a higher drying creep.9.14,9.125,9.152 It
should be remembered that the various cementitious materials have different rates of
hydration and therefore of gain of strength while the concrete is under load. The rate
of gain of strength affects creep; this was referred to earlier in this section.
An example of the influence of hydration on creep is offered by tests of Buil and
Acker,9.150 who found that silica fume has no effect on basic creep but significantly
reduces the drying creep. The explanation is likely to lie in the fact that the hydration
reactions of silica fume reduce the amount of water available for movement out of the
gel. Generally, because of the long-term hydration, and therefore increase in strength
under sustained load, of concrete containing fly ash or ground granulated blastfurnace
slag, the long-term rate of creep is reduced in such concrete.
Creep of concrete made with expansive cement is larger than when Portland
cement only is included in the mix.9.156
Water-reducing and set-retarding admixtures have been found to increase the basic
creep in many, but not all, cases.9.134,9.135 There are indications that lignosulfonate-based
admixtures lead to a larger increase than carboxylic-acid-based admixtures.9.71 For
drying creep, no reliable pattern of influence of these admixtures has been
established.9.71 The same situation exists with respect to superplasticizers.9.71 From this
rather unsatisfactory situation, it follows that, if creep is of importance in a given
structure, the influence of any admixture to be used should be carefully checked.
There are some general comments that should be made about differences in creep
reported by various researchers. In a number of investigations, the differences in creep
reported are of about the same magnitude as the scatter of results for any one set of
tests. It is not reasonable, therefore, to accept these differences as significant and they
cannot be used as a basis for prediction. Tests involving actual materials are necessary.
These tests, which must be performed under conditions expected to apply in service,
can be of short duration. Extrapolation using the expressions discussed on p. 470 can
be used to estimate long-term creep.
Reverting to the relation between creep and the stress/strength ratio, we can note
that because, for a given mix, strength and modulus of elasticity are related to one
another, creep and modulus of elasticity are also related. Figure 9.34 shows

497
experimental values of creep at any time t, against the ratio of the modulus of
elasticity at the time t to the modulus at the time of application of load;9.118 the ages at
which the load was applied and at which creep was determined varied widely, but one
mix only was used. The modulus at the time of application of the load gives an
indication of the strength at that time, and the increase in the modulus reflects the
duration of the load.

Fig. 9.34. Relation between creep at any time t and ratio of the modulus of
elasticity of concrete at time t to the modulus at the time of application of the
load; various concretes, ages at loading, and periods under load9.118
Influence of ambient relative humidity
One of the most important external factors influencing creep is the relative humidity
of the air surrounding the concrete. Taking a broad view, we can say that, for a given
concrete, creep is higher the lower the relative humidity. This is illustrated in Fig.
9.35 for specimens cured at a relative humidity of 100 per cent and then loaded and
exposed to different humidities. Such treatment results in a greatly varying shrinkage
occurring in the different specimens during the early stages after the application of the
sustained load. The rates of creep during that period vary correspondingly but, at later
ages, the rates seem to be close to one another. Thus, drying while under load
enhances creep of concrete, i.e. induces the additional drying creep (cf. Fig. 9.28).
The influence of relative humidity is much smaller, or absent, in the case of
specimens which have reached hygral equilibrium with the surrounding medium prior
to the application of the load9.117 (Fig. 9.35). Thus, in reality, it is not the relative

498
humidity that influences creep but the process of drying, i.e. the occurrence of drying
creep.

Fig. 9.35. Creep of concrete cured in fog for 28 days, then loaded and stored at
different relative humidities9.24
Drying creep may be related to, or influenced by, the tensile stress induced in the
outer part of a concrete specimen by restrained shrinkage and the resultant
cracking.9.149 The compressive stress arising from an applied compressive load cancels
out this cracking.9.148 Consequently, the actual shrinkage of a loaded specimen is larger
than the measured shrinkage of a specimen which has undergone surface cracking.
The approach of considering creep and shrinkage to be additive, therefore, assumes
too small a value of shrinkage: the difference between this assumed shrinkage and the
actual shrinkage in a loaded specimen represents drying creep. This hypothesis has
not, however, been confirmed by tests on mortar9.145 in which a large drying creep was
observed in the absence of shrinkage cracking of unloaded companion specimens.
Day and Illston9.154 also found that very small specimens of hydrated cement paste
undergo drying creep and concluded that drying creep is an intrinsic property of
hydrated cement paste.
Bažant and Xi9.157 suggested that, rather than drying creep, there exists stress-
induced shrinkage caused by local movement of water between capillary pores and
gel pores. However, until convincing evidence is available, the concept of drying
creep as defined in Fig. 9.28 should be retained.
At this stage, it is appropriate to note that concrete which exhibits high shrinkage
shows generally also a high creep.9.14 This does not mean that the two phenomena are
due to the same cause, but they may both be linked to the same aspect of the structure
of hydrated cement paste. It should not be forgotten that concrete cured and loaded at
a constant relative humidity exhibits creep, and that creep produces no significant loss
of water from the concrete to the surrounding medium;9.120,9.121 nor is there any gain in

499
mass during creep recovery.9.121 (A small increase in mass occasionally observed
during the period of creep or creep recovery may be due to carbonation.)
A further indication of the interrelation between shrinkage and creep is given
in Fig. 9.36. Specimens which had been loaded for 600 days and then unloaded and
allowed to recover their creep exhibited, on subsequent immersion in water, swelling
proportional to the stress which had been removed over two years previously. The
residual deformation after swelling shows a similar proportionality.

Fig. 9.36. Relation between original sustained stress and: (a) expansion in water,
and (b) residual deformation of concrete9.113
Figure 9.37 shows time deformation of loaded specimens stored alternately in
water and in air with a relative humidity of 50 per cent. The ordinates represent the
change in deformation from that existing after 600 days under load in air. It can be
seen that, while in water, the loaded specimens show creep relative to the swelling of
the unloaded specimen, but in air the change in deformation of all specimens is the
same. The increase in creep on immersion of this old concrete in water may be due to
the breaking of some of the bonds formed during the period of drying (cf.
p. 443). Figure 9.38 shows the data of Fig. 9.37 plotted as a deformation relative to
the deformation of the unloaded specimen. A practical conclusion from these
observations is that alternating wetting and drying increases the magnitude of creep,

500
so that results of laboratory tests may underestimate the creep under normal weather
conditions.

Fig. 9.37. Time deformation of concrete subjected to different stresses and stored
alternately in water and in air at a relative humidity of 50 per cent.9.14 Strains at
origin of time (after 600 days under load in air):

501
 Fig. 9.38. Time deformation of loaded specimens of Fig. 9.37 plotted relative to
the strain of the unloaded specimen9.14
Creep has been found to decrease with an increase in the size of the specimen. This
may be due to the effects of shrinkage and to the fact that creep at the surface occurs
under conditions of drying and is, therefore, greater than within the core of the
specimen where the conditions approximate to mass curing. Even if, with time, drying
reaches the core, it will have hydrated extensively and reached a higher strength,
which leads to lower creep. In sealed concrete, no size effects can be present.
The size effect can best be expressed in terms of the volume/surface ratio of the
concrete member; the relation is shown in Fig. 9.39. It can be seen that the actual
shape of the specimen is of even lesser importance than in the case of shrinkage. Also,
the decrease in creep with an increase in size is smaller than in the case of shrinkage
(cf. Fig. 9.21). But the rates of gain of creep and of shrinkage are the same, indicating
that both phenomena are the same function of the volume/surface ratio. These data
apply to shrinkage and creep at 50 per cent relative humidity.9.53

502
 Fig. 9.39. Relation between ratio of creep to elastic strain and volume/surface
ratio9.53
Other influences
The influence of temperature on creep is of interest in prestressed concrete nuclear
pressure vessels as well as in other types of structures, e.g. bridges. The rate of creep
increases with temperature up to about 70 °C (160 °F) when, for a 1:7 mix with a
water/cement ratio of 0.6, it is approximately 3.5 times higher than at 21 °C (70 °F).
Between 70 °C (160 °F) and 96 °C (205 °F) the rate drops off to 1.7 times the rate at
21 °C (70 °F).9.116 These differences in rate persist at least for 15 months under
load. Figure 9.40 illustrates the progress of creep. This behaviour is believed to be due
to desorption of water from the surface of the gel so that gradually the gel itself
becomes the sole phase subject to molecular diffusion and shear flow; consequently
the rate of creep decreases. It is also possible that a part of the increase in the creep of
concrete loaded at elevated temperatures may be due to the lower strength of concrete
at high temperatures9.147 (see p. 361).

503
 Fig. 9.40. Relation between creep and time under load for concretes stored at
different temperatures (stress/strength ratio of 0.70)9.116
As far as low temperatures are concerned, freezing produces a higher initial rate of
creep but it quickly drops to zero.9.137 At temperatures between –10 and –30 °C (14 and
–22 °F). creep is about one-half of the creep at 20 °C (68 °F).9.155
Creep of concrete over a wide range of temperatures is shown in Fig. 9.41.9.136

504
 Fig. 9.41. Influence of temperature on rate of creep9.136
Most of the test data on creep have been obtained under a sustained constant stress
but sometimes the actual load alternates between some limits. It has been found that
an alternating load, with a given mean stress/strength ratio, leads to a larger time-
dependent deformation than a static load corresponding to the same stress/strength
ratio.9.139 This is illustrated in Fig. 9.42 for the case when the alternating load varied
between a stress/strength ratio 0.35 and 0.05 while the static load represented a
stress/strength ratio of 0.35. The same figure shows also the deformation under a
mean stress/strength ratio of 0.35 (varying between 0.45 and 0.25): the deformation is
higher still. The deformation under cyclic loading is caused probably by the same
mechanism as creep under a static load so that the use of the term ‘creep’ in both
cases may be justified. It seems that cycling results in a higher rate of creep at early
ages and also leads to a larger long-term value.9.140 Thus, use of creep data from static
tests may underestimate creep when the load is cyclic.

505
 Fig. 9.42. Creep under alternating and static loading
The preceding discussion referred to uniaxial compression but creep also occurs in
other loading situations, and information about creep behaviour under these
conditions is especially helpful in establishing the nature of creep and in some design
problems. Unfortunately, experimental data are limited, and in many cases
quantitative evaluation and comparison with the behaviour in compression are not
possible. For this reason, no more than broad qualitative statements will be made.
Creep of mass concrete in uniaxial tension is 20 to 30 per cent higher than under a
compressive stress of equal magnitude. The difference depends upon age at loading
and may be as high as 100 per cent for storage at a relative humidity of 50 per cent for
concrete loaded at early ages. However, contradictory evidence also exists9.101 so that
reliable statements about creep in tension cannot be made. The shape of the creep–
time curves in tension is broadly similar to that in compression, but the decrease in the
rate of creep with time is much less pronounced in the former case because the
increase in strength with age is lower. Drying enhances creep in tension just as it does
in compression. In direct tension, time failure occurs in a manner similar to uniaxial
compression, but the critical stress/strength ratio is probably only 0.7.9.158
Creep occurs under torsional loading, and is affected by stress, water/cement ratio
and ambient relative humidity in qualitatively the same manner as creep in

506
compression. The creep–time curve is also of the same shape.9.119 The ratio of creep to
elastic deformation in torsion was found to be the same as for compressive loading.9.138
Under uniaxial compression, creep occurs not only in the axial direction but also in
the normal directions. This is referred to as lateral creep. The resulting creep
Poisson’s ratio was considered on p. 423. From the fact that there is lateral creep
induced by an axial stress, it follows that, under multiaxial stress, in any direction
there is creep due to the stress applied in that direction and also creep due to the
Poisson’s ratio effect of creep strains in the two normal directions. There is
evidence9.45 that the superposition of creep strains due to each stress separately is not
valid, so that creep under multiaxial stress cannot be simply predicted from uniaxial
creep measurements. Specifically, creep under multiaxial compression is less than
under a uniaxial compression of the same magnitude in the given direction (Fig. 9.43).
But even under hydrostatic compression there is considerable creep.

Fig. 9.43. Typical creep–time curves for concrete under triaxial compression
Relation between creep and time
Creep is usually determined by measuring the change with time in the strain of a
specimen subjected to a constant stress and stored under appropriate conditions.
ASTM C 512-02 describes a spring-loaded frame which maintains a constant load on
a concrete test cylinder despite any change in its length. However, for comparative
tests on concrete with untried aggregates or admixtures, an even simpler test
apparatus can be used9.141 (Fig. 9.44). Here, the load has to be adjusted from time to
time, its value being determined by a dynamometer in series with the concrete
specimens.

507
 Fig. 9.44. A simple test rig for the determination of creep of concrete under an
approximately constant stress9.141
The apparatus of Fig. 9.44 can be used for accelerated creep tests by immersion in
water at a temperature of between 45 and 65 °C. As mentioned earlier, a higher
temperature leads to a higher creep so that, after 7 days, any difference between an
unknown concrete and a reference concrete can be easily detected. This accelerated
creep appears to be linearly related to the 100-day creep at normal temperature for a
wide range of mixes and aggregates,9.141 as shown in Fig. 9.45.

508
 Fig. 9.45. Relation between creep in the accelerated 7-day test at a higher
temperature and 100-day creep at normal temperature for various concrete
mixes9.141
Creep continues for a very long time, if not indefinitely, the longest determination
to date indicating that a small increase in creep takes place after as long as 30
years9.24 (Fig. 9.46); the tests were then discontinued because of interference by
carbonation of the specimens. The rate of creep decreases, however, at a continuous
rate, and it is generally assumed that creep tends to a limiting value after an infinite
time under load; this has not, however, been proved.

509
 Fig. 9.46. Range of creep–time curves for different concretes stored at various
relative humidities9.24
Figure 9.46 gives long-term measurements of Troxell et al.,9.24 and it can be seen
that, if creep after 1 year under load is taken as unity, then the average values of creep
at later ages are:
1.14 after 2 years
1.20 after 5 years
1.26 after 10 years
1.33 after 20 years, and
1.36 after 30 years.
These values show that ultimate creep may be in excess of 1.36 times the one-year
creep, although for calculation purposes it is often assumed that the 30-year creep
represents the ultimate creep.
Numerous mathematical expressions relating creep and time have been suggested.
One of the most convenient is the hyperbolic expression, introduced by Ross9.122 and by
Lorman.9.31 Ross expresses creep c after time t under load as

When t = ∞, then c = 1/b, i.e. 1/b is the limiting value of creep. The
symbols a and b represent constants determined from experimental results: by
plotting t/c against t, a straight line of slope b is obtained, and the intercept of
the t/c axis is equal to a. The straight line should be drawn so as to pass through the
points at later ages, there being generally some deviation from the straight line during
the early period after the application of the load.

510
 ACI 209R-929.80 uses a modified Ross expression, the main difference being the
application of a power exponent of 0.6 to time t. ACI 209R-92 also offers values of
coefficients to allow for various factors influencing creep.
The U.S. Bureau of Reclamation, which has made an extensive study of creep of
concrete in dams, where only basic creep occurs, has found that creep can be
represented by an expression of the type
c = F(K) loge (t+ 1)
where K = age at which the load is applied,
F(K) = a function representing the rate of creep deformation with time, and
t = time under load, in days.
F(K) is obtained from a plot on semi-logarithmic paper.
Sometimes, values of creep per unit stress are given, usually in units of 10–6 per
MPa; this is known as specific creep or unit creep. Creep can also be expressed as a
ratio of creep to the initial elastic deformation; this ratio is known as creep
coefficient or characteristic creep. The merit of this approach is that it takes into
account the elastic properties of aggregate, which influence creep and the elastic
deformation of concrete in a similar manner.
All embracing, but complex, expressions for creep have been developed by Bažant
and co-workers, who published also a somewhat simplified, but not simple, version of
creep prediction expressions.9.146
The variety of creep expressions may seem bewildering but a reliable prediction of
creep of any concrete under any conditions is not possible. Short-term tests, say of a
28-day duration under load, are necessary. Extrapolation is then possible. It has been
found9.142 that for periods under load up to five years, the power expression seems best
to fit experimental data for basic creep and, for basic-plus-drying creep, a logarithmic-
power function appears most appropriate. For the majority of concretes, regardless of
the water/cement ratio or the type of aggregate, specific creep at the age of t days (t >
28), ct, can be related to specific creep after 28 days under load, c28, by the expressions:
basic creep: ct = c28 × 0.50t0.21
total creep: ct = c28 × (–6.19 + 2.15 loge t)0.38
where ct = long-term specific creep in 10–6 per MPa.
Nature of creep
From Fig. 9.30, it is apparent that creep and creep recovery are related phenomena,
but their nature is far from clear. The fact that creep is partly reversible suggests that
it may consist of a partly reversible visco-elastic movement (consisting of a purely
viscous phase and a purely elastic phase) and possibly also a non-reversible plastic
deformation.
An elastic deformation is always recoverable on unloading. A plastic deformation
is never recoverable, can be time-dependent, and there is no proportionality between
plastic strain and the applied stress, or between stress and rate of strain.
A viscous deformation is never recoverable on unloading, is always time-dependent,
and there is always proportionality between the rate of viscous strain and the applied
stress, and hence between stress and strain at a given time.9.129 These various types of
deformation can be summarized as shown in Table 9.6.

511
 Table 9.6. Types of Deformation

A possible treatment of the observed partial recovery of creep is by the principle of

superposition of strains, developed by McHenry.9.126 This states that the strains
produced in concrete at any time t by a stress increment applied at any time t0 are
independent of the effects of any stress applied either earlier or later than t0. The stress
increment is understood to mean either a compressive or a tensile stress, i.e. also a
relief of load. It follows then that, if the compressive stress on a specimen is removed
at age t1, the resulting creep recovery will be the same as the creep of a similar
specimen subjected to the same compressive stress at the age t1. Figure 9.47 illustrates
this statement, and it can be seen that the creep recovery is represented by the
difference between the actual strain at any time and the strain that would exist at the
same time, had the specimen continued to be subjected to the original compressive
stress.

Fig. 9.47. Example of McHenry’s principle of superposition of strains9.126

Figure 9.48 shows a comparison of actual and computed strains (the computed
values being in reality the difference between two experimental curves) for sealed
concrete, i.e. subject to basic creep only.9.127 It appears that, in all cases, the actual
strain after the removal of load is higher than the residual strain predicted by the
principle of superposition. Thus, actual creep is less than expected. Similar error is
found when the principle is applied to specimens under a variable stress.9.107 It seems
then that the principle of superposition does not fully satisfy the phenomena of creep
and creep recovery.

512
 Fig. 9.48. Comparison of measured and computed strains on the basis of
McHenry’s principle of superposition9.127
The principle of superposition of strains is, nevertheless, a convenient working
assumption. It implies that creep is a delayed elastic phenomenon in which full
recovery is generally impeded by the progressive hydration of the cement. Because
the properties of old concrete change only very little with age, creep of concrete
subjected to sustained loading at the age of several years would be expected to be
fully reversible, but this has not been verified experimentally. It should be noted that
the principle of superposition leads to a tolerable error under mass-curing conditions,
i.e. when only basic creep occurs. When drying creep is present, the error is large in
that the creep recovery is grossly overestimated.
The problem of the nature of creep is still controversial9.128 and cannot be discussed
here in full. The locus of creep is the hydrated cement paste, and creep is related to
internal movement of adsorbed or intracrystalline water, i.e. to internal seepage.
Glucklich’s tests9.132 have shown that concrete from which all evaporable water has
been removed exhibits practically no creep. However, the changes in the creep
behaviour of concrete at high temperatures suggest that, at that stage, the water ceases
to play a role and the gel itself becomes subjected to creep deformation.
Because creep can take place in mass concrete, it follows that seepage of water to
the outside of concrete is not essential to the progress of basic creep, although such a
process may well take place in drying creep. However, internal seepage of water from
the adsorbed layers to voids such as capillary voids is possible. An indirect evidence

513
of the role of such voids is given by the relation between creep and the strength of the
hydrated cement paste: it would appear that creep is a function of the relative amount
of the unfilled space, and it can be speculated that it is the voids in the gel that govern
both strength and creep; in the latter case, the voids may be related to seepage. The
volume of voids is, of course, a function of the water/cement ratio and is affected by
the degree of hydration.
We should remember that capillary voids do not remain full even against full
hydrostatic pressure of a water bath. Thus, internal seepage is possible under any
storage conditions. The fact that creep of non-shrinking specimens is independent of
the ambient relative humidity would indicate that the fundamental cause of creep ‘in
air’ and ‘in water’ is the same.
The creep–time curve shows a definite decrease in its slope, and the question arises
whether this signifies a change, possibly a gradual one, in the mechanism of creep. It
is conceivable that the slope decreases with the same mechanism continuing
throughout, but it is reasonable to imagine that, after many years under load, the
thickness of the adsorbed water layers could be reduced so far that no further
reduction can take place under the same stress, and yet creep after as many as 30
years has been recorded. It is, therefore, probable that the slow, long-term part of
creep is due to causes other than seepage but the deformation can develop only in the
presence of some evaporable water. This would suggest viscous flow or sliding
between the gel particles. Such mechanisms are compatible with the influence of
temperature on creep, and can explain also the largely irreversible character of long-
term creep.
Observations on creep under cyclic loading, and especially on the temperature rise
within concrete under such loading, have led to a modified hypothesis of creep. As
already mentioned, creep under a cyclic stress is increased compared with creep under
a static stress equal to the mean cyclic stress.9.140 This increased creep is largely
irrecoverable and consists of accelerated creep due to increased viscous sliding of the
gel particles, and of increased creep due to a limited amount of microcracking at very
early stages of the creep process. Other experimental data on creep in tension and in
compression9.143 suggest that the behaviour is best explained by a combination of the
seepage and viscous shear theories of creep.
Generally, the role of microcracking is small and, excepting cyclic creep, is
probably limited to concrete loaded at very early ages, and loaded to high
stress/strength ratios in excess of 0.6.
Having said all this, we have to admit that the exact mechanism of creep remains
uncertain.
Effects of creep
Creep affects strains and deflections and often also stress distribution, but the effects
vary with the type of structure.9.130
Creep of plain concrete does not per se affect the strength, although
under very high stresses creep hastens the approach of the limiting strain at which
failure takes place; this applies only when the sustained load is above 85 or 90 per
cent of the rapidly applied static ultimate load.9.115 Under a low sustained stress, the
volume of concrete decreases (as the creep Poisson’s ratio is less than 0.5) and this
would be expected to increase the strength of the concrete. However, this effect is
probably small.

514
 The influence of creep on the behaviour and strength of reinforced and prestressed
concrete structures is fully discussed in ref. 9.84. Here, it may be useful to mention
that, in reinforced concrete columns, creep results in a gradual transfer of load from
the concrete to the reinforcement. Once the steel yields, any increase in load is taken
by the concrete, so that the full strength of both the steel and the concrete is developed
before failure takes place – a fact recognized by the design formulae. However, in
eccentrically loaded columns, creep increases the deflection and can lead to buckling.
In statically indeterminate structures, creep may relieve stress concentrations induced
by shrinkage, temperature changes, or movement of supports. In all concrete
structures, creep reduces internal stresses due to non-uniform shrinkage, so that there
is a reduction in cracking. In calculating creep effects in structures, it is important to
realize that the actual time-dependent deformation is not the ‘free’ creep of concrete
but a value modified by the quantity and position of reinforcement.
On the other hand, in mass concrete, creep in itself may be a cause of cracking
when a restrained concrete mass undergoes a cycle of temperature change due to the
development of the heat of hydration and subsequent cooling. A compressive stress is
induced by the rapid rise in temperature in the interior of the concrete mass. This
stress is low because the modulus of elasticity of very young concrete is low. The
strength of very young concrete is also low so that its creep is high; this relieves the
compressive stress, and the remaining compression disappears as soon as some
cooling has taken place. On further cooling of concrete, tensile stresses develop and,
because the rate of creep is reduced with age, cracking may occur even before the
temperature has dropped to the initial (placing) value (see Fig. 9.49). For this reason,
the rise in temperature in the interior of a large concrete mass must be controlled (see
p. 395).

515
 Fig. 9.49. Stress in concrete subjected to a temperature cycle at a constant
length9.131
Creep can also lead to an excessive deflection of structural members and cause
other serviceability problems, especially in high-rise buildings and long bridges.
The loss of prestress due to creep is well known and, indeed, accounts for the
failure of the original attempts at prestressing.
The effects of creep may thus be harmful but, on the whole, creep, unlike
shrinkage, is beneficial in relieving stress concentrations and has contributed very
considerably to the success of concrete as a structural material. Rational design
methods allowing for creep in various types of structures have been developed.9.112
References
9.1. R. E. PHILLEO, Comparison of results of three methods for determining
Young’s modulus of elasticity of concrete, J. Amer. Concr. Inst., 51, pp.
461–9 (Jan. 1955).
9.2. H. LOSSIER, Cements with controlled expansions and their applications to
prestressed concrete, The Structural Engineer, 24, No. 10, pp. 505–34
(1946).
9.3. M. POLIVKA, Factors influencing expansion of expansive cement
concretes. Klein Symp. on Expansive Cement, ACI SP-38, pp. 239–50
(Detroit, Michigan, 1973).
9.4. M. POLIVKA and C. WILLSON, Properties of shrinkage-compensating
concretes, Klein Symp. on Expansive Cement, ACI SP-38, pp. 227–37
(Detroit, Michigan, 1973).
9.5. L. W. TELLER, Elastic properties, ASTM Sp. Tech. Publ. No. 169, pp. 94–103
(1956).
9.6. J. J. SHIDELER, Lightweight aggregate concrete for structural use, J. Amer.
Concr. Inst., 54, pp. 299–328 (Oct. 1957).
9.7. P. KLIEGER, Early high-strength concrete for prestressing. Proc. World
Conference on Prestressed Concrete, pp. A5-1–14, (San Francisco, 1957).
9.8. M. KOKUBU, Use of expansive components for concrete in Japan. Klein Symp.
on Expansive Cement, ACI SP-38, pp. 353–78 (Detroit, Michigan, 1973).
9.9. T. TAKABAYASHI, Comparison of dynamic Young’s modulus and static
Young’s modulus for concrete, RILEM Int. Symp. on Non-destructive
Testing of Materials and Structures 1, pp. 34–44, (1954).
9.10. J. BIJEN and G. VAN DER WEGEN, Swelling of concrete in deep
seawater, Durability of Concrete, Ed. V. M. Malhotra, ACI SP-145, pp.
389–407 (Detroit, Michigan, 1994).
9.11. B. W. SHACKLOCK and P. W. KEENE, A comparison of the compressive and
flexural strengths of concrete with and without entrained air, Civil
Engineering (London), pp. 77–80 (Jan. 1959).
9.12. R. JONES, Testing concrete by an ultrasonic pulse technique, D.S.I.R. Road
Research Technical Paper No. 34 (London, HMSO, 1955).
9.13. M. A. SWAYZE, Early concrete volume changes and their control, J. Amer.
Concr. Inst., 38, pp. 425–40 (April 1942).

516
 9.14. R. L’HERMITE, Volume changes of concrete, Proc. 4th Int. Symp. on the
Chemistry of Cement, Washington DC, pp. 659–94 (1960).
9.15. W. LERCH, Plastic shrinkage, J. Amer. Concr. Inst., 53, pp. 797–802 (Feb.
1957).
9.16. D. W. HOBBS, Influence of specimen geometry upon weight change and
shrinkage of air-dried concrete specimens, Mag. Concr. Res., 29, No. 99, pp.
70–80 (1977).
9.17. H. E. DAVIS, Autogenous volume changes of concrete, Proc. ASTM., 40, pp.
1103–10 (1940).
9.18. T. C. POWERS, Causes and control of volume change, J. Portl. Cem. Assoc.
Research and Development Laboratories, 1, No. 1, pp. 29–39 (Jan. 1959).
9.19. F. M. LEA, The Chemistry of Cement and Concrete (London, Arnold, 1970).
9.20. J. D. BERNAL, J. W. JEFFERY and H. F. W. TAYLOR, Crystallographic research on
the hydration of Portland cement: A first report on investigations in
progress, Mag. Concr. Res., 3, No. 11, pp. 49–54 (1952).
9.21. J. D. BERNAL, The structures of cement hydration compounds, Proc. 3rd Int.
Symp. on the Chemistry of Cement, London, pp. 216–36 (1952).
9.22. F. M. LEA, Cement research: Retrospect and prospect, Proc. 4th Int. Symp.
on the Chemistry of Cement, Washington DC, pp. 5–8 (1960).
9.23. G. PICKETT, Effect of aggregate on shrinkage of concrete and hypothesis
concerning shrinkage. J. Amer. Concr. Inst., 52, pp. 581–90 (Jan. 1956).
9.24. G. E. TROXELL, J. M. RAPHAEL and R. E. DAVIS, Long-time creep and shrinkage
tests of plain and reinforced concrete, Proc. ASTM., 58, pp. 1101–20 (1958).
9.25. B. W. SHACKLOCK and P. W. KEENE, The effect of mix proportions and testing
conditions on drying shrinkage and moisture movement of concrete, Cement
Concr. Assoc. Tech. Report TRA/266 (London, June 1957).
9.26. M. A. SWAYZE, Discussion on: Volume changes of concrete. Proc. 4th Int.
Symp. on the Chemistry of Cement, Washington DC, pp. 700–2 (1960).
9.27. G. PICKETT, Effect of gypsum content and other factors on shrinkage of
concrete prisms, J. Amer. Concr. Inst., 44, pp. 149–75 (Oct. 1947).
9.28. W. LERCH, The influence of gypsum on the hydration and properties of
portland cement pastes, Proc. ASTM., 46, pp. 1252–92 (1946).
9.29. P. W. KEENE, The effect of air-entrainment on the shrinkage of concrete
stored in laboratory air, Cement Concr. Assoc. Tech. Report
TRA/331 (London, Jan. 1960).
9.30. J. J. SHIDELER, Calcium chloride in concrete, J. Amer. Concr. Inst., 48, pp.
537–59 (March 1952).
9.31. W. R. LORMAN, The theory of concrete creep, Proc. ASTM., 40, pp. 1082–102
(1940).
9.32. A. D. ROSS, Shape, size, and shrinkage, Concrete and Constructional
Engineering, pp. 193–9 (London, Aug. 1944).
9.33. R. L’HERMITE, J. CHEFDEVILLE and J. J. GRIEU, Nouvelle contribution à l’étude
du retrait des ciments, Annales de l’Institut Technique du Bâtiment et de
Travaux Publics No. 106. Liants Hydrauliques No. 5 (Dec. 1949).

517
 9.34. J. W. GALLOWAY and H. M. HARDING, Elastic moduli of a lean and a pavement
quality concrete under uniaxial tension and compression, Materials and
Structures, 9, No. 49, pp. 13–18 (1976).
9.35. A. M. NEVILLE, Discussion on: Effect of aggregate on shrinkage of concrete
and hypothesis concerning shrinkage, J. Amer. Concr. Inst., 52, Part 2, pp.
1380–1 (Dec. 1956).
9.36. P. T. WANG, S. P. SHAH and A. E. NAAMAN, Stress–strain curves of normal and
lightweight concrete in compression, J. Amer. Concr. Inst., 75, pp. 603–11
(Nov. 1978).
9.37. G. J. VERBECK, Carbonation of hydrated portland cement, ASTM. Sp. Tech.
Publ. No. 205, pp. 17–36 (1958).
9.38. J. J. SHIDELER, Investigation of the moisture-volume stability of concrete
masonry units, Portl. Cem. Assoc. Development Bull. D.3 (March 1955).
9.39. R. N. SWAMY and A. K. BANDYOPADHYAY, The elastic properties of structural
lightweight concrete, Proc. Inst. Civ. Engrs., Part 2, 59, pp. 381–94 (Sept.
1975).
9.40. A. M. NEVILLE, Shrinkage and creep in concrete, Structural Concrete, 1, No.
2, pp. 49–85 (London, March 1962).
9.41. E. W. BENNETT and D. R. LOAT, Shrinkage and creep of concrete as affected
by the fineness of Portland cement, Mag. Concr. Res., 22, No. 71, pp. 69–78
(1970).
9.42. S. P. SHAH and G. WINTER, Inelastic behaviour and fracture of concrete, Symp.
on Causes, Mechanism, and Control of Cracking in Concrete, ACI SP-20,
pp. 5–28 (Detroit, Michigan, 1968).
9.43. A. M. NEVILLE, Some problems in inelasticity of concrete and its behaviour
under sustained loading, Structural Concrete, 3, No. 4, pp. 261–8 (London,
1966).
9.44. P. DESAYI and S. KRISHNAN, Equation for the stress–strain curve of concrete, J.
Amer. Concr. Inst., 61, pp. 345–50 (March 1964).
9.45. K. S. GOPALAKRISHNAN, A. M. NEVILLE and A. GHALI, Creep Poisson’s ratio of
concrete under multiaxial compression, J. Amer. Concr. Inst., 66, pp. 1008–
20 (Dec. 1969).
9.46. I. E. HOUK, O. E. BORGE and D. L. HOUGHTON, Studies of autogenous volume
change in concrete for Dworshak Dam, J. Amer. Concr. Inst., 66, pp. 560–8
(July 1969).
9.47. D. RAVINA and R. SHALON, Plastic shrinkage cracking. J. Amer. Concr.
Inst., 65, pp. 282–92 (April 1968).
9.48. S. T. A. ÖDMAN, Effects of variations in volume, surface area exposed to
drying, and composition of concrete on shrinkage, RILEM/CEMBUREAU
Int. Colloquium on the Shrinkage of Hydraulic Concretes, 1, 20 pp. (Madrid,
1968).
9.49. T. W. REICHARD, Creep and drying shrinkage of lightweight and normal
weight concretes. Nat. Bur. Stand. Monograph, 74, (Washington DC, March
1964).

518
 9.50. K. MATHER, High strength, high density concrete, J. Amer. Concr. Inst., 62,
No. 8, pp. 951–62 (1965).
9.51. S. E. PIHLAJAVAARA, Notes on the drying of concrete, Reports, Series 3, No. 79
(Helsinki, The State Institute for Technical Research, 1963).
9.52. T. C. HANSEN, Effect of wind on creep and drying shrinkage of hardened
cement mortar and concrete, ASTM Mat. Res. & Stand., 6, pp. 16–19 (Jan.
1966).
9.53. T. C. HANSEN and A. H. MATTOCK, The influence of size and shape of member
on the shrinkage and creep of concrete, J. Amer. Concr. Inst., 63, pp. 267–
90 (Feb. 1966).
9.54. J. W. KELLY, Cracks in concrete – the causes and cures, Concrete
Construction, 9, pp. 89–93 (April 1964).
9.55. R. G. L’HERMITE, Quelques problèmes mal connus de la technologie du
béton, Il Cemento, 75, No. 3, pp. 231–46 (1978).
9.56. S. E. PIHLAJAVAARA, On practical estimation of moisture content of drying
concrete structures, Il Cemento, 73, No. 3, pp. 129–38 (1976).
9.57. S. POPOVICS, Verification of relationships between mechanical properties of
concrete-like materials, Materials and Structures, 8, No. 45, pp. 183–91
(1975).
9.58. S. E. PIHLAJAVAARA, Carbonation – an important effect on the surfaces of
cement-based materials, RILEM/ASTM/CIB Symp. on Evaluation of the
Performance of External Surfaces of Buildings; Paper No. 9, 9 pp.
(Otaniemi, Finland, Aug. 1977).
9.59. N. J. GARDNER, P. L. SAU and M. S. CHEUNG, Strength development and
durability of concrete, ACI Materials Journal, 85, No. 6, pp. 529–36 (1988).
9.60. S. HARSH, Z. SHEN and D. DARWIN, Strain-rate sensitive behavior of cement
paste and mortar in compression, ACI Materials Journal, 87, No. 5, pp.
508–16 (1990).
9.61. Z.-H. GUO and X.-Q. ZHANG, Investigation of complete stress–deformation
curves for concrete in tension, ACI Materials Journal, 84, No. 4, pp. 278–85
(1987).
9.62. W. S. NAJJAR and K. C. HOVER, Neutron radiography for microcrack studies of
concrete cylinders subjected to concentric and eccentric compressive
loads, ACI Materials Journal, 86, No. 4, pp. 354–9 (1989).
9.63. F. DE LARRARD, E. SAINT-DIZIER and C. BOULAY, Comportement post-rupture de
béton à hautes ou très hautes performances armé en compression, Bulletin
Liaison Laboratoires Ponts et Chaussées, 179, pp. 11–20 (May–June 1992).
9.64. N. H. OLSEN, H. KRENCHEL and S. P. SHAH, Mechanical properties of high
strength concrete, IABSE Symposium, Concrete Structures for the Future,
Paris – Versailles, pp. 395–400 (1987).
9.65. W. F. CHEN, Concrete plasticity: macro- and microapproaches, Int. Journal of
Mechanical Sciences, 35, No. 12, pp. 1097–109 (1993).
9.66. M. M. SMADI and F. O. SLATE, Microcracking of high and normal strength
concretes under short- and long-term loadings, ACI Materials Journal, 86,
No. 2, pp. 117–27 (1989).

519
 9.67. D. J. CARREIRA and K.-H. CHU, Stress–strain relationship for plain concrete in
compression, ACI Journal, 82, No. 6, pp. 797–804 (1985).
9.68. K. J. BASTGEN and V. HERMANN, Experience made in determining the static
modulus of elasticity of concrete, Materials and Structures, 10, No. 60, pp.
357–64 (1977).
9.69. P.-C. AÏTCIN, M. S. CHEUNG and V. K. SHAH, Strength development of concrete
cured under arctic sea conditions, in Temperature Effects on Concrete,
ASTM Sp. Tech. Publ. No. 858, pp. 3–20 (Philadelphia, Pa, 1983).
9.70. F. D. LYDON and R. V. BALENDRAN, Some observations on elastic properties of
plain concrete, Cement and Concrete Research, 16, No. 3, pp. 314–24
(1986).
9.71. J. J. BROOKS and A. NEVILLE, Creep and shrinkage of concrete as affected by
admixtures and cement replacement materials, in Creep and Shrinkage of
Concrete: Effect of Materials and Environment, ACI SP-135, pp. 19–36
(Detroit, Michigan, 1992).
9.72. W. HANSEN and J. A. ALMUDAIHEEM, Ultimate drying shrinkage of concrete –
influence of major parameters, ACI Materials Journal, 84, No. 3, pp. 217–
23 (1987).
9.73. J. BARON, Les retraits de la pâte de ciment, in Le Béton Hydraulique –
Connaissance et Pratique, Eds J. Baron and R. Santeray, pp. 485–501
(Paris, Presses de l’École Nationale des Ponts et Chaussées, 1982).
9.74. J.-M. TORRENTI et al., Contraintes initiales dans le béton, Bulletin Liaison
Ponts et Chaussées, 158, pp. 39–44 (Nov.–Dec. 1988).
9.75. R. MENSI, P. ACKER and A. ATTOLOU, Séchage du béton: analyse et
modélisation, Materials and Structures, 21, No. 121, pp. 3–12 (1988).
9.76. M. SHOYA, Drying shrinkage and moisture loss of super plasticizer admixed
concrete of low water cement ratio, Transactions of the Japan Concrete
Institute, II – 5, pp. 103–10 (1979).
9.77. J. J. BROOKS, Influence of mix proportions, plasticizers and superplasticizers
on creep and drying shrinkage of concrete, Mag. Concr. Res. 41, No. 148,
pp. 145–54 (1989).
9.78. R. W. CARLSON and T. J. READING, Model study of shrinkage cracking in
concrete building walls, ACI Structural Journal, 85, No. 4, pp. 395–404
(1988).
9.79. M. GRZYBOWSKI and S. P. SHAH, Shrinkage cracking of fiber reinforced
concrete, ACI Materials Journal, 87, No. 2, pp. 138–48 (1990).
9.80. ACI 209R-92, Prediction of creep, shrinkage, and temperature effects in
concrete structures, ACI Manual of Concrete Practice Part 1: Materials and
General Properties of Concrete, 47 pp. (Detroit, Michigan, 1994).
9.81. E. J. SELLEVOLD, Shrinkage of concrete: effect of binder composition and
aggregate volume fraction from 0 to 60%, Nordic Concrete Research,
Publication No. 11, pp. 139–52 (Oslo, The Nordic Concrete Federation, Feb.
1992).
9.82. R. D. GAYNOR, R. C. MEININGER and T. S. KHAN, Effect of temperature and
delivery time on concrete proportions, in Temperature Effects on Concrete,
ASTM Sp. Tech. Publ. No. 858, pp. 68–87 (Philadelphia, Pa, 1983).

520
 9.83. J. A. ALMUDAIHEEM and W. HANSEN, Effect of specimen size and shape on
drying shrinkage, ACI Materials Journal, 84, No. 2, pp. 130–4 (1987).
9.84. A. M. NEVILLE, W. H. DILGER and J. J. BROOKS, Creep of Plain and Structural
Concrete, 361 pp. (London, Construction Press, Longman Group, 1983).
9.85. G. C. HOFF and K. MATHER, A look at Type K shrinkage-compensating
cement production and specifications, Cedric Willson Symposium on
Expansive Cement, ACI SP-64, pp. 153–80 (Detroit, Michigan, 1977).
9.86. R. W. CUSICK and C. E. KESLER, Behavior of shrinkage-compensating
concretes suitable for use in bridge decks, Cedric Willson Symposium on
Expansive Cement, ACI SP-64, pp. 293–301 (Detroit, Michigan, 1977).
9.87. B. MATHER, Curing of concrete, Lewis H. Tuthill International Symposium on
Concrete and Concrete Construction, ACI SP-104, pp. 145–59 (Detroit,
Michigan, 1987).
9.88. E. TAZAWA and S. MIYAZAWA, Autogenous shrinkage of concrete and its
importance in concrete, in Creep and Shrinkage in Concrete, Eds Z. P.
Bažant and I. Carol, Proc. 5th International RILEM Symposium, pp. 159–68
(London, E & FN Spon, 1993).
9.89. C. LOBO and M. D. COHEN, Hydration of Type K expansive cement paste and
the effect of silica fume: II. Pore solution analysis and proposed hydration
mechanism, Cement and Concrete Research, 23, No. 1, pp. 104–14 (1993).
9.90. M. D. COHEN, J. OLEK and B. MATHER, Silica fume improves expansive cement
concrete, Concrete International, 13, No. 3, pp. 31–7 (1991).
9.91. ACI 223-93, Standard practice for the use of shrinkage-compensating
concrete, ACI Manual of Concrete Practice Part 1: Materials and General
Properties of Concrete, 26 pp. (Detroit, Michigan, 1994).
9.92. YAN FU, S. A. SHEIKH and R. D. HOOTON, Microstructure of highly expansive
cement paste, ACI Materials Journal, 91, No. 1, pp. 46–54 (1994).
9.93. G. GIACCIO et al., High-strength concretes incorporating different coarse
aggregates, ACI Materials Journal, 89, No. 3, pp. 242–6 (1992).
9.94. F. A. OLUOKUN, Prediction of concrete tensile strength from its compressive
strength: evaluation of existing relations for normal weight concrete, ACI
Materials Journal, 88, No. 3, pp. 302–9 (1991).
9.95. M. KAKIZAKI et al., Effect of Mixing Method on Mechanical Properties and
Pore Structure of Ultra High-Strength Concrete, Katri Report No. 90, 19 pp.
(Tokyo, Kajima Corporation, 1992) (and also in ACI SP-132,
CANMET/ACI, 1992).
9.96. ACI 517.2R-87, Revised 1992, Accelerated curing of concrete at
atmospheric pressure – state of the art, ACI Manual of Concrete Practice
Part 5: Masonry, Precast Concrete, Special Processes, 17 pp. (Detroit,
Michigan, 1994).
9.97. ACI 305R-91, Hot weather concreting, ACI Manual of Concrete Practice
Part 2: Construction Practices and Inspection Pavements, 20 pp. (Detroit,
Michigan, 1994).
9.98. ACI 318-02 Building code requirements for structural concrete, ACI Manual
of Concrete Practice Part 3: Use of Concrete in Buildings – Design,
Specifications, and Related Topics, 443 pp.

521
 9.99. ACI 363R-92, State-of-the-art report on high-strength concrete, ACI Manual
of Concrete Practice Part 1: Materials and General Properties of Concrete,
55 pp. (Detroit, Michigan, 1994).
9.100. E. K. ATTIOGBE and D. DARWIN, Submicrocracking in cement paste and
mortar, ACI Materials Journal, 84, No. 6, pp. 491–500 (1987).
9.101. A. YONEKURA, M. KUSAKA and S. TANAKA, Tensile creep of early age concrete
with compressive stress history, Cement Association of Japan Review, pp.
158–61 (1988).
9.102. Y. H. LOO and G. D. BASE, Variation of creep Poisson’s ratio with stress in
concrete under short-term uniaxial compression, Mag. Concr. Res., 42, No.
151, pp. 67–73 (1990).
9.103. M. D. COHEN, J. OLEK and W. L. DOLCH, Mechanism of plastic shrinkage
cracking in portland cement and portland cement–silica fume paste and
mortar, Cement and Concrete Research, 20, No. 1, pp. 103–19 (1990).
9.104. Y. F. HOUST, Influence of shrinkage on carbonation shrinkage kinetics of
hydrated cement paste, in Creep and Shrinkage of Concrete, Eds. Z. P.
Bažant and I. Carol, Proc. 5th Int. RILEM Symp, Barcelona, pp. 121–6
(London, E & FN Spon, 1993).
9.105. A. NEVILLE, Whither expansive cement?, Concrete International, 16, No. 9,
pp. 34–5 (1994).
9.106. R. N. SWAMY, Shrinkage characteristics of ultra-rapid-hardening
cement, Indian Concrete J., 48, No. 4, pp. 127–31 (1974).
9.107. A. D. ROSS, Creep of concrete under variable stress, J. Amer. Concr. Inst., 54,
pp. 739–58 (March 1958).
9.108. A. M. NEVILLE, Creep recovery of mortars made with different cements, J.
Amer. Concr. Inst., 56, pp. 167–74 (Aug. 1959).
9.109. A. M. NEVILLE, Creep of concrete as a function of its cement paste
content, Mag. Concr. Res., 16, No. 46, pp. 21–30 (1964).
9.110. S. E. RUTLEDGE and A. M. NEVILLE, Influence of cement paste content on creep
of lightweight aggregate concrete, Mag. Concr. Res., 18, No. 55, pp. 69–74
(1966).
9.111. H. RÜSCH, K. KORDINA and H. HILSDORF, Der Einfluss des mineralogischen
Charakters der Zuschläge auf das Kriechen von Beton, Deutscher Ausschuss
für Stahlbeton, No. 146, pp. 19–133 (Berlin, 1963).
9.112. A. M. NEVILLE, Creep of Concrete: plain, and prestressed (North-Holland,
Amsterdam, 1970).
9.113. A. M. NEVILLE, The relation between creep of concrete and the stress–
strength ratio, Applied Scientific Research, Section A, 9, pp. 285–92 (The
Hague, 1960).
9.114. L. L. YUE and L. TAERWE, Creep recovery of plain concrete and its
mathematical modelling, Magazine of Concrete Research, 44, No. 161, pp.
281–90 (1992).
9.115. A. M. NEVILLE, Rôle of cement in the creep of mortar, J. Amer. Concr.
Inst., 55, pp. 963–84 (March 1959).

522
 9.116. K. W. NASSER and A. M. NEVILLE, Creep of concrete at elevated
temperatures, J. Amer. Concr. Inst., 62, pp. 1567–79 (Dec. 1965).
9.117. A. M. NEVILLE, Tests on the influence of the properties of cement on the
creep of mortar, RILEM Bull. No. 4, pp. 5–17 (Oct. 1959).
9.118. U.S. BUREAU OF RECLAMATION, A 10-year study of creep properties of
concrete, Concrete Laboratory Report No. SP-38 (Denver, Colorado, 28 July
1953).
9.119. B. LE CAMUS, Recherches expérimentales sur la déformation du béton et du
béton armé, Comptes Rendues des Recherches des Laboratoires du Bâtiment
et des Travaux Publics (Pairs, 1945–46).
9.120. G. A. MANEY, Concrete under sustained working loads; evidence that
shrinkage dominates time yield, Proc. ASTM., 41, pp. 1021–30 (1941).
9.121. A. M. NEVILLE, Recovery of creep and observations on the mechanism of
creep of concrete, Applied Scientific Research, Section A, 9, pp. 71–84 (The
Hague, 1960).
9.122. A. D. ROSS, Concrete creep data, The Structural Engineer, 15, pp. 314–26
(London, 1937).
9.123. A. M. NEVILLE and H. W. KENINGTON, Creep of aluminous cement
concrete, Proc. 4th Int. Symp. on the Chemistry of Cement, Washington DC,
pp. 703–8 (1960).
9.124. A. M. NEVILLE, The influence of cement on creep of concrete in mortar, J.
Prestressed Concrete Inst., pp. 12–18 (Gainesville, Florida, March 1958).
9.125. A. D. ROSS, The creep of Portland blast-furnace cement concrete, J. Inst. Civ.
Engrs., pp. 43–52 (London, Feb. 1938).
9.126. D. MCHENRY, A new aspect of creep in concrete and its application to
design, Proc. ASTM., 43, pp. 1069–84 (1943).
9.127. U.S. BUREAU OF RECLAMATION, Supplemental Report – 5-year creep and strain
recovery of concrete for Hungry Horse Dam, Concrete Laboratory Report
No. C-179A (Denver, Colorado, 6 Jan. 1959).
9.128. A. M. NEVILLE, Theories of creep in concrete. J. Amer. Conor. Inst., 52, pp.
47–60 (Sept. 1955).
9.129. T. C. HANSEN, Creep of concrete – a discussion of some fundamental
problems, Swedish Cement and Concrete Research Inst., Bull, No. 33 (Sept.
1958).
9.130. A. M. NEVILLE, Non-elastic deformations in concrete structures, J. New
Zealand Inst. E., 12, pp. 114–20 (April 1957).
9.131. R. E. DAVIS, H. E. DAVIS and E. H. BROWN, Plastic flow and volume change of
concrete, Proc. ASTM, 37, Part II, pp. 317–30 (1937).
9.132. J. GLUCKLICH, Creep mechanism in cement mortar, J. Amer. Conor. Inst., 59,
pp. 923–48 (July 1962).
9.133. A. M. NEVILLE, M. M. STAUNTON and G. M. BONN, A study of the relation
between creep and the gain of strength of concrete, Symp. on Structure of
Portland Cement Paste and Concrete, Highw. Res. Bd, Special Report No.
90, pp. 186–203 (Washington DC, 1966).

523
 9.134. B. B. HOPE, A. M. NEVILLE and A. GURUSWAMI, Influence of admixtures on
creep of concrete containing normal weight aggregate, RILEM Int. Symp. on
Admixtures for Mortar and Concrete, pp. 17–32 (Brussels, Sept. 1967).
9.135. E. L. JESSOP, M. A. WARD and A. M. NEVILLE, Influence of water reducing and
set retarding admixtures on creep of lightweight aggregate concrete, RILEM
Int. Symp. on Admixtures for Mortar and Concrete, pp. 35–46 (Brussels,
Sept. 1967).
9.136. J. C. MARÉCHAL, Le fluage du béton en fonction de la température, Materials
and Structures, 2, No. 8, pp. 111–15 (1969).
9.137. R. JOHANSEN and C. H. BEST, Creep of concrete with and without ice in the
system, RILEM Bull. No. 16, pp. 47–57 (Paris, Sept. 1962).
9.138. H. LAMBOTTE, Le fluage du béton en torsion, RILEM Bull. No. 17, pp. 3–12
(Paris, Dec. 1962).
9.139. A. M. NEVILLE and C. P. WHALEY, Non-elastic deformation of concrete under
cyclic compression, Mag. Conor. Res., 25, No. 84, pp. 145–54 (1973).
9.140. A. M. NEVILLE and G. HIRST, Mechanism of cyclic creep of concrete, Douglas
McHenry International Symposium on Concrete and Concrete Structures,
ACI SP-55 pp. 83–101 (Detroit, Michigan, 1978).
9.141. A. M. NEVILLE and W. Z. LISZKA, Accelerated determination of creep of
lightweight aggregate concrete, Civil Engineering, 68, pp. 515–19 (London,
June 1973).
9.142. J. J. BROOKS and A. M. NEVILLE, Predicting long-term creep and shrinkage
from short-term tests, Mag. Conor. Res., 30, No. 103, pp. 51–61 (1978).
9.143. J. J. BROOKS and A. M. NEVILLE, A comparison of creep, elasticity and strength
of concrete in tension and in compression, Mag. Concr. Res., 29, No. 100,
pp. 131–41 (1977).
9.144. M. D. LUTHER and W. HANSEN, Comparison of creep and shrinkage of high-
strength silica fume concretes with fly ash concretes of similar strengths,
in Fly Ash, Silica Fume, Slag, and Natural Pozzolans in Concretes, Proc.
3rd International Conference, Trondheim, Norway, Vol. 1, ACI SP-114, pp.
573–91 (Detroit, Michigan, 1989).
9.145. A. BENAÏSSA, P. MORLIER and C. VIGUIER, Fluage et retrait du béton de
sable, Materials and Structures, 26, No. 160, pp. 333–9 (1993).
9.146. Z. P. BAžANT et al., Improved prediction model for time-dependent
deformations of concrete: Part 6 – simplified code-type
formulation, Materials and Structures, 25, No. 148, pp. 219–23 (1992).
9.147. W. P. S. DIAS, G. A. KHOURY and P. J. E. SULLIVAN, The thermal and structural
effects of elevated temperatures on the basic creep of hardened cement
paste, Materials and Structures, 23, No. 138, pp. 418–425 (1990).
9.148. P. ROSSI and P. ACKER, A new approach to the basic creep and relaxation of
concrete, Cement and Concrete Research, 18, No. 5, pp. 799–803 (1988).
9.149. F. H. WITTMANN and P. E. ROELFSTRA, Total deformation of loaded drying
concrete, Cement and Concrete Research, 10, No. 5, pp. 601–10 (1980).
9.150. M. BUIL and P. ACKER, Creep of silica fume concrete, Cement and Concrete
Research, 15, No. 3, pp. 463–7 (1985).

524
 9.151. E. TAZAWA, A. YONEKURA and S. TANAKA, Drying shrinkage and creep of
concrete containing granulated blast furnace slag, in Fly Ash, Silica Fume,
Slag, and Natural Pozzolans in Concretes, Proc. 3rd International
Conference, Trondheim, Norway, Vol. 2, ACI SP-114, pp. 1325–43 (Detroit,
Michigan, 1989).
9.152. J.-C. CHERN and Y.-W. CHAN, Deformations of concrete made with blast-
furnace slag cement and ordinary portland cement, ACI Materials
Journal, 86, No. 4, pp. 372–82 (1989).
9.153. K. W. NASSER and A. A. AL-MANASEER, Creep of concrete containing fly ash
and superplasticizer at different stress/strength ratios, ACI Journal, 83, No. 4,
pp. 668–73 (1986).
9.154. R. L. DAY and J. M. ILLSTON, The effect of rate of drying on the
drying/wetting behaviour of hardened cement paste, Cement and Concrete
Research, 13, No. 1, pp. 7–17 (1983).
9.155. F. H. TURNER, Concrete and Cryogenics – Part 1, Concrete, 14, No. 5, pp. 39–
40 (1980).
9.156. H. G. RUSSELL, Performance of shrinkage-compensating concrete in
slabs, Research and Development Bulletin, RD057.01D, 12 pp. (Skokie, Ill.,
Portland Cement Association, 1978).
9.157. Z. P. BAžANT and YUNPING XI, Drying creep of concrete: constitutive model
and new experiments separating its mechanisms, Materials and
Structures, 27, No. 165, pp. 3–15 (1994).
9.158. H. T. SHKOUKANI, Behaviour of concrete under concentric and eccentric
tensile loading, Darmstadt Concrete, 4, pp. 113–232 (1989).
9.159. A. NEVILLE, Neville on Concrete: An Examination of Issues in Concrete
Practice, Second Edition (Book Surge, LLC, www.createspace.com, 2006).
9.160. CIRIA, Early age thermal crack control in concrete, Report C 660, 2007.

525
Chapter 10. Durability of concrete

It is essential that every concrete structure should continue to perform its intended
functions, that is maintain its required strength and serviceability, during the specified
or traditionally expected service life. It follows that concrete must be able to
withstand the processes of deterioration to which it can be expected to be exposed.
Such concrete is said to be durable.
It is worth adding that durability does not mean an indefinite life, nor does it mean
withstanding any action on concrete. Moreover, it is nowadays realized, although it
was not so in the past, that, in many situations, routine maintenance of concrete is
required;10.68 an example of maintenance procedures is given by Carter.10.72
The fact that durability has not been considered in this book up to now could be
interpreted to mean that this topic is of lesser importance than other properties of
concrete, notably strength. This is not so, and, indeed, in many situations, durability is
of paramount importance. Nevertheless, until recently, developments in cement and
concrete technology have concentrated on achieving higher and higher strengths (see
p. 334). There was an assumption that ‘strong concrete is durable concrete’, the only
special considerations being the effects of alternating freezing and thawing and some
forms of chemical attack. It is now known that, for many conditions of exposure of
concrete structures, both strength and durability have to be considered explicitly at the
design stage. The emphasis is on the word ‘both’ because it would be a mistake to
replace overemphasis on strength by overemphasis on durability. Durability
requirements for a service life of 50 and 100 years are given in BS 8500-1 : 2006.
This chapter considers various aspects of durability. Two special topics, the effects
of freezing and thawing, including the action of de-icing agents, and chloride attack,
are the subject matter of Chapter 11.
Causes of inadequate durability
Inadequate durability manifests itself by deterioration which can be due either to
external factors or to internal causes within the concrete itself. The various actions can
be physical, chemical, or mechanical. Mechanical damage is caused by impact
(considered on p. 345), abrasion, erosion or cavitation; the last three are discussed
towards the end of the present chapter. The chemical causes of deterioration include
the alkali-silica and alkali-carbonate reactions which are also discussed in this chapter.
External chemical attack occurs mainly through the action of aggressive ions, such as
chlorides, sulfates, or of carbon dioxide, as well as many natural or industrial liquids
and gases. The damaging action can be of various kinds and can be direct or indirect.
Physical causes of deterioration include the effects of high temperature or of the
differences in thermal expansion of aggregate and of the hardened cement paste
(discussed in Chapter 8). An important cause of damage is alternating freezing and
thawing of concrete and the associated action of de-icing salts; these topics are
discussed in Chapter 11.
It should be observed that the physical and chemical processes of deterioration can
act in a synergistic manner. The various factors affecting the durability of concrete are
the subject matter of the present chapter. At this stage, it is worth noting that
deterioration of concrete is rarely due to one isolated cause: concrete can often be
satisfactory despite some undesirable features but, with an additional adverse factor,

526
damage will occur. For this reason, it is sometimes difficult to assign deterioration to
a particular factor, but the quality of concrete, in the broad sense of the word, though
with a special reference to permeability, nearly always enters the picture. Indeed, with
the exception of mechanical damage, all the adverse influences on durability involve
the transport of fluids through the concrete. For this reason, consideration of
durability requires an understanding of the phenomena involved.
Transport of fluids in concrete
There are three fluids principally relevant to durability which can enter concrete:
water, pure or carrying aggressive ions, carbon dioxide and oxygen. They can move
through the concrete in different ways, but all transport depends primarily on the
structure of the hydrated cement paste. As stated earlier, durability of concrete largely
depends on the ease with which fluids, both liquids and gases, can enter into, and
move through, the concrete; this is commonly referred to as permeability of concrete.
Strictly speaking, permeability refers to flow through a porous medium. Now, the
movement of the various fluids through concrete takes place not only by flow through
the porous system but also by diffusion and sorption, so that our concern is really
with penetrability of concrete. Nevertheless, the commonly accepted term
‘permeability’ will be used for the overall movement of fluids into and through
concrete except where, for clarity, distinctions between the various types of flow need
to be made.
Towards the end of 2010 a paper10.142 was published describing and discussing the
transport phenomena involved in evaluating the measurement of chlorides in concrete:
a total of 11 phenomena are listed. This paper is of great value in helping to
understand the transport phenomena involved. Alas, in a practical situation concrete
fluids do not follow a single mode of transport, nor is a single type of ion involved. I
know that wick action is involved in the old-fashioned oil lamp but, even there, there
is a confusion between the UK and US in the name of the fluid involved: paraffin and
kerosene.
These remarks do not detract from the value of the paper: they simply illustrate the
need to understand the phenomena involved which have to be applied to a given
situation.
Influence of the pore system
The aspect of the structure of hardened cement paste relevant to permeability is the
nature of the pore system within the bulk of the hardened cement paste and also in the
zone near the interface between the cement paste and the aggregate. The interface
zone occupies as much as one-third to one-half of the total volume of hardened
cement paste in concrete and is known to have a different microstructure from the
bulk of the hardened cement paste. The interface is also the locus of early
microcracking. For these reasons, the interface zone can be expected significantly to
contribute to the permeability of concrete.10.44 However, Larbi10.49 found that, despite the
higher porosity of the interface zone, the permeability of concrete is controlled by the
bulk of the hardened cement paste, which is the only continuous phase in concrete.
Support to Larbi’s view is lent by the fact that the permeability of hardened cement
paste is not lower than that of concrete made with a similar cement paste. However, of
relevance in concrete is also the fact that any movement of fluids has to follow a path
made longer and more tortuous by the presence of aggregate, which also reduces the
effective area for flow. Thus, the significance of the interface zone with respect to

527
permeability remains uncertain. Even more generally, it has to be admitted that the
relationship between permeability and the pore structure of hardened cement paste is,
at best, qualitative.10.97
The pores relevant to permeability are those with a diameter of at least 120 or 160
nm. These pores have to be continuous. Pores which are ineffective with respect to
flow, that is to permeability, include, in addition to discontinuous pores, those which
contain adsorbed water and those which have a narrow entrance, even if the pores
themselves are large (cf. Fig. 6.16).
Aggregate can also contain pores, but these are usually discontinuous. Moreover,
aggregate particles are enveloped by the cement paste so that the pores in aggregate
do not contribute to the permeability of concrete. The same applies to discrete air
voids, such as entrained-air bubbles (see p. 547). In addition, the concrete as a whole
contains voids caused by incomplete compaction or by trapped bleed water. These
voids may occupy between a fraction of one per cent and 10 per cent of the volume of
the concrete, the latter figure representing a highly honeycombed concrete of very low
strength. Such concrete or concrete with leaking joints should not be made, and it will
not be further discussed.
Flow, diffusion, and sorption
Because of the existence of pores of different kinds, some of which contribute to
permeability and some of which do not do so, it is important to distinguish between
porosity and permeability. Porosity is a measure of the proportion of the total volume
of concrete occupied by pores, and is usually expressed in per cent. If the porosity is
high and the pores are interconnected, they contribute to the transport of fluids
through concrete so that its permeability is also high. On the other hand, if the pores
are discontinuous or otherwise ineffective with respect to transport, then the
permeability of the concrete is low, even if its porosity is high.
Porosity can be measured by mercury intrusion; this topic was referred to on p. 283,
and is comprehensively treated by Cook and Hover.10.46 Other fluids can also be used.
An indication of porosity can be obtained from the measurement of absorption of
concrete, which is considered on p. 489.
As far as the ease of movement of fluids through concrete is concerned, thus far
loosely referred to as permeability, three mechanisms should be
distinguished. Permeability refers to flow under a pressure differential. Diffusion is
the process in which a fluid moves under a differential in concentration; the relevant
property of concrete is diffusivity. Gases can diffuse through water-filled space or
through air-filled space but, in the former case, the process is 104 to 105 times slower
than in the latter.
Sorption is the result of capillary movement in the pores in concrete which are
open to the ambient medium. It follows that capillary suction can take place only in
partially dry concrete; there is no sorption of water in completely dry concrete or in
saturated concrete.
Because the penetrability of concrete is described in the literature in varying terms,
it is important to present briefly the relevant mathematical expressions and to state
clearly the units of measurement. A comprehensive discussion of various aspects of
permeability is presented in ref. 10.96.
Coefficient of permeability

528
Flow in capillary pores in saturated concrete follows Darcy’s law for laminar flow
through a porous medium:

where dq/dt = rate of flow of water in m3/s,

A = cross-sectional area of the sample in m2,
Δh = drop in hydraulic head through the sample, measured in m,
L = thickness of the sample in m,
η = dynamic viscosity of the fluid in N s/m2,
ρ = density of the fluid in kg/m3, and
g = acceleration due to gravity.
The coefficient K′ is then expressed in metres to the power 2, and represents
the intrinsic permeability of the material, independently of the fluid involved.
As the fluid involved is generally water, we can put:

The coefficient K is then expressed in metres per second and is referred to as

the coefficient of permeability of concrete, it being understood that it refers to water at
room temperature. The last qualification arises from the fact that the viscosity of
water changes with temperature. The flow equation can thus be written as:

and when a steady state of flow dq/dt has been reached, K is determined direct.
Diffusion
As stated earlier, when the transport of a gas or a vapour through concrete is the result
of a concentration gradient, and not of a pressure differential, diffusion takes place.
As far as the diffusion of gases is concerned, carbon dioxide and oxygen are of
primary interest: the former leads to carbonation of hydrated cement paste, and the
latter makes possible the progress of corrosion of embedded steel. The first of these
mechanisms of deterioration is discussed later in this chapter; corrosion is considered
in Chapter 11. At this stage, it is useful to note that the diffusivity coefficient of a gas
is inversely proportional to the square root of its molar mass,10.130 so that, for instance,
oxygen diffuses, theoretically, 1.17 times faster than carbon dioxide. This relation
makes it possible to calculate the diffusion coefficient of one gas from experimental
data on another gas.
Diffusion coefficient
The diffusion equation applicable to water vapour and air can be expressed by Fick’s
first law as:

529
where dc/dL = concentration gradient in kg/m4 or moles/m4,
D = diffusion coefficient in m2/s,
J = mass transport rate in kg/m2 s (or moles/m2 s), and
L = thickness of the sample in metres.
Even though diffusion takes place only through the pores, the values
of J and D refer to the cross-section of the concrete sample; thus, D is, in reality,
the effective diffusion coefficient.
The diffusion coefficient of a gas can be determined experimentally under a
steady-state system, with two sides of a concrete specimen being exposed, each to a
different pure gas: the mass of the gases on the side opposite to that where they were
originally present is then determined. The pressure on each side of the specimen
should be the same because the driving force in diffusion is the difference in molar
concentration and not a pressure differential.
Diffusion through air and water
Papadakis et al.10.130 present expressions for the effective diffusion coefficient of carbon
dioxide as a function of the relative humidity of the air and of the porosity of
hardened cement paste or of the compressive strength of concrete. The diffusion
through water is 4 orders of magnitude slower than through air. It should be noted that
the diffusion coefficient changes with age because the pore system in concrete
changes with time, especially when hydration of cement continues.
Oxygen diffusion through concrete is strongly affected by moist
curing,10.96 prolonged curing reducing the diffusion coefficient by a factor of about 6.
The moisture condition of the concrete under test also has a large influence because
water in the pores significantly reduces the diffusion. By way of illustration, the
oxygen diffusion coefficient of well-cured concrete, conditioned at a relative humidity
of 55 per cent, is less than 5 × 10–8 m2/s for high-quality concrete, and more than 50 ×
10–8 m2/s for poor-quality concrete.10.96
Movement of water vapour through concrete can occur as a result of a humidity
differential on its two opposed sides.10.12 The relative humidity on the two sides of the
concrete has to be known because an increase in relative humidity decreases the air-
filled pore space available for diffusion. It follows then that, if the moist side is, for
instance, saturated, an increase in the relative humidity of the dry side reduces the
vapour permeability. Water vapour transmission of concrete is generally affected in a
similar manner to air permeability.
In addition to the diffusion of gases, ions of aggressive character, notably chlorides
and sulfates, move by diffusion in the pore water. It is in the pore water that reactions
with hydrated cement paste take place so that ionic diffusion is of importance with
respect to sulfate attack of concrete and chloride attack of embedded steel. Ionic
diffusion is most effective when the pores in the hardened cement paste are saturated,
but it can also take place in partially saturated concrete.
Like permeability, diffusion is lower at lower water/cement ratios, but the
influence of the water/cement ratio on diffusion is much smaller than on permeability.
Absorption
The volume of pore space in concrete, as distinct from the ease with which a fluid can
penetrate it, is measured by absorption; the two quantities are not necessarily related.

530
Absorption is usually measured by drying a specimen to a constant mass, immersing it
in water, and measuring the increase in mass as a percentage of dry mass. Various
procedures can be used, and widely different results are obtained as shown in Table
10.1. One reason for this variation in the values of absorption is that, at one extreme,
drying at ordinary temperature may be ineffective in removing all the water; on the
other hand, drying at high temperatures may remove some of the combined water.
Absorption cannot, therefore, be used as a measure of quality of concrete, but most
good concretes have an absorption well below 10 per cent by mass; if the volume
occupied by water is to be calculated, an allowance for the difference in the specific
gravity of water and of concrete needs to be made.

Table 10.1. Values of Absorption of Concrete Determined in Various Ways10.7

An absorption test on several small portions of concrete is prescribed by ASTM C

642-06; drying at 100 to 110 °C (212 to 230 °F) and immersion in water at 21 °C
(70 °F) for at least 48 hours are used. The requirements of BS 1881-122 : 1983 are
similar except that the test is performed on whole core specimens.
Absorption tests are not used frequently except for routine quality control of
precast products such as paving flags, slabs or kerb (curb) units. The absorption of
sawn-off small test specimens, dried for 72 hours at 105 °C, and then immersed in
water for 30 minutes and for 24 hours, is determined.
Surface absorption tests
For practical purposes, it is the absorption characteristics of the outer zone of concrete
(which offers protection to reinforcement) that are of greatest interest. For that reason,
tests measuring the surface absorption have been developed.
A test to determine the initial surface absorption is prescribed in BS 1881-5 : 1984
(withdrawn). In essence, the rate of absorption of water by the surface zone of
concrete is determined during a prescribed period (ranging between 10 minutes and 1

531
hour) under a head of 200 mm (8 in.) of water: this head is only slightly greater than
that which would be caused by driving rain. The rate of initial surface absorption is
expressed in millilitres per square metre per second.
Initial absorption after 10 minutes greater than 0.50 ml/m2 per second would be
considered high, and smaller than 0.25 ml/m2 per second, low. Corresponding values
after 2 hours are, respectively, greater than 0.15 ml/m2 per second and smaller than
0.07 ml/m2 per second.10.96
A shortcoming of the initial surface absorption test is that the flow of water
through the concrete is not uni-directional. To remedy this, several modified tests
have been proposed but none has gained general acceptance.
The mass of water which is absorbed by concrete during the test depends on the
pre-existing moisture content. For this reason, the results of the initial surface
absorption test cannot be readily interpreted unless the concrete has been conditioned
to a known hygrometric state prior to the tests. This requirement cannot be satisfied in
in-situ concrete. In consequence, a low value of the initial surface absorption may be
due either to the inherent low absorption characteristics of the concrete tested or else
to the fact that the pores in poor-quality concrete are already full of water.
Bearing the above limitation in mind, the initial surface absorption test can be used
to compare the effectiveness of curing of the outer zone of concrete.
A test which gives some measure of the ease with which water or air enters
concrete in situ was developed by Figg.10.22 A small hole is drilled and sealed with
silicone rubber. This plug is pierced by a hypodermic needle connected to a vacuum
pump, and pressure in the system is reduced by a given amount. The time required for
air to permeate through the concrete and increase the pressure in the cavity to a
specified value is an indication of the air ‘permeability’ of the concrete. Another
model of the apparatus makes it possible to assess the water ‘permeability’ of
concrete by measuring the time for a given volume of water to enter the
concrete.10.22 Several modifications of the Figg apparatus have been developed.10.96
It should be pointed out that the term ‘permeability’ is not really valid because the
output of the Figg tests is not directly related to the coefficient of permeability as
properly defined. Nevertheless, the tests are useful for comparative purposes.
Sorptivity
Because of the difficulties associated with the absorption tests, on the one hand, and,
on the other, because permeability tests measure the response of concrete to pressure,
which is rarely the driving force of fluids entering concrete, there is a need for another
type of test. Such a test measures the rate of absorption of water by capillary suction
of unsaturated concrete placed in contact with water; no head of water exists.
Essentially, the sorptivity test determines the rate of capillary-rise absorption by a
concrete prism which rests on small supports in a manner such that only the lowest 2
to 5 mm of the prism is submerged. The increase in the mass of the prism with time is
recorded.
It has been shown10.98 that there exists a relation of the form
i = St0.5
where i = increase in mass since the beginning of the testing per unit of cross-
sectional area in contact with water, divided by the density of water. Working in
metric units i can be expressed in mm

532
 t = time, measured in minutes, at which the mass is determined, and
S = sorptivity in mm/min0.5.
In practice, it is easier to measure the value of i as a rise in the water level in the
concrete, which manifests itself by a darker colour. In such a case, i is measured direct
in millimetres. If sorptivity is to be expressed in consistent SI units, then the following
conversion can be used:

1 mm/min0.5 = 1.29 × 10–4 m/s0.5.

In the test, several measurements are taken over a period of up to 4 hours, and a
straight line is fitted to the plot of the increase in mass, or the rise of the water front,
versus the square root of time. The point of origin (and possibly also the very early
readings) is ignored because there is a small increase in mass at the instant when the
open surface pores in the lowest 2 to 5 mm of the prism first become submerged
(see Fig. 10.1).

Fig. 10.1. Example of relation between increase in mass of water per unit area
and time used to calculate sorptivity
Some typical values of sorptivity are: 0.09 mm/min0.5 for concrete with a
water/cement ratio of 0.4, and 0.17 mm/min0.5 at a water/cement ratio ratio of 0.6;
these should not be considered as anything more than examples.
As in the initial surface absorption test, the higher the moisture content of the
concrete the lower the measured sorptivity so that, if possible, the specimen should be
conditioned at 105 °C prior to testing; alternatively, the hygral state of the specimen
should be established.
Water permeability of concrete
The principles of flow of water through concrete under pressure were discussed on
p. 486 in terms of flow through a porous body. Some more specific features of the
permeability of concrete will now be considered.
First of all, we can note that the hardened cement paste is composed of particles
connected over only a small fraction of their total surface. For this reason, a part of
the water is within the field of force of the solid phase, i.e. it is adsorbed. This water
has a high viscosity but is, nevertheless, mobile and takes part in the flow.10.2 As

533
already stated, the permeability of concrete is not a simple function of its porosity, but
depends also on the size, distribution, shape, tortuosity, and continuity of the pores.
Thus, although the cement gel has a porosity of 28 per cent, its permeability10.3 is only
about 7 × 10–16 m/s. This is due to the extremely fine texture of hardened cement paste:
the pores and the solid particles are very small and numerous, whereas, in rocks, the
pores, though fewer in number, are much larger and lead to a higher permeability. For
the same reason, water can flow more easily through the capillary pores than through
the much smaller gel pores: the cement paste as a whole is 20 to 100 times more
permeable than the gel itself.10.3 It follows that the permeability of hardened cement
paste is controlled by its capillary porosity. The relation between these two quantities
is shown in Fig. 10.2. For comparison, Table 10.2 lists the water/cement ratio of
pastes having the same permeability as some common rocks.10.3 It is interesting to see
that the permeability of granite is about the same as that of mature cement paste with
a water/cement ratio of 0.7, i.e. not of high quality.

Fig. 10.2. Relation between permeability and capillary porosity of cement paste10.3

Table 10.2. Comparison Between Permeabilities of Rocks and Cement Pastes10.3

534
 The permeability of cement paste varies with the progress of hydration. In a fresh
paste, the flow of water is controlled by the size, shape, and concentration of the
original cement particles. With the progress of hydration, the permeability decreases
rapidly because the gross volume of gel (including the gel pores) is approximately 2.1
times the volume of the unhydrated cement, so that the gel gradually fills some of the
original water-filled space. In a mature paste, the permeability depends on size, shape,
and concentration of the gel particles and on whether or not the capillaries have
become discontinuous.10.4 Table 10.3 gives values of the coefficient of
permeability10.5 at different ages for a cement paste with a water/cement ratio of 0.7.
The reduction in the coefficient of permeability is faster the lower the water/cement
ratio of the paste, so that there is little reduction after wet curing for a period of:10.21
7 days when the water/cement ratio is 0.45
28 days when the water/cement ratio is 0.60
90 days when the water/cement ratio is 0.70.

Table 10.3. Reduction in Permeability of Cement Paste (Water/Cement Ratio =

0.7) with the Progress of Hydration10.5

For cement pastes hydrated to the same degree, the permeability is lower the
higher the cement content of the paste, i.e. the lower the water/cement ratio. Figure
10.3 shows values obtained for pastes in which 93 per cent of the cement has
hydrated.10.5 The slope of the line is considerably lower for pastes with water/cement
ratios below about 0.6, i.e. pastes in which some capillaries have become segmented
(see p. 32). From Fig. 10.3 it can be seen that a reduction of water/cement ratio from,
say, 0.7 to 0.3 lowers the coefficient of permeability by 3 orders of magnitude. The

535
same reduction occurs in a paste with a water/cement ratio of 0.7 between the ages of
7 days and one year.

Fig. 10.3. Relation between permeability and water/cement ratio for mature
cement pastes10.5 (93 per cent of cement hydrated)
In concrete, the value of the coefficient of permeability decreases very
substantially with a decrease in the water/cement ratio: over the range of
water/cement ratios of 0.75 to 0.26, the coefficient decreases by up to 4 orders of
magnitude,10.51 and over the range of 0.75 to 0.45, by 2 orders of magnitude.
Specifically, at a water/cement ratio of 0.75, the coefficient of permeability
is typically 10–10 m/s, and this would be considered to represent concrete with a high
permeability. At a water/cement ratio of 0.45, the coefficient is typically 10–11 or 10–
12 m/s; permeabilities of an order of magnitude lower than the last value are considered

to represent concretes with a very low permeability.

In this connection, it is useful to refer again to Fig. 10.3 which applies to mature
cement pastes. There is a large increase in the permeability at water/cement ratios in
excess of about 0.4. In the vicinity of this water/cement ratio, the capillaries become
segmented so that there is a substantial difference in permeability between mature
cement pastes with a water/cement ratio below 0.4 and those with higher
water/cement ratios. This difference has implications for ingress of aggressive ions
into concrete. Permeability of concrete is also of interest in relation to water-tightness
of liquid-retaining and some other structures, and also with reference to the problem
of hydrostatic pressure in the interior of dams. Furthermore, ingress of moisture into
concrete affects its thermal insulation properties (see pp. 376 and 708).
Increasing the wet-curing period of concrete with a very high water/cement ratio
from 1 day to 7 days was found10.51 to reduce water permeability by a factor of 5.

536
 The permeability of concrete is affected also by the properties of cement. For the
same water/cement ratio, coarse cement tends to produce a hardened cement paste
with a higher porosity than a finer cement.10.5 The compound composition of the
cement affects permeability in so far as it influences the rate of hydration, but the
ultimate porosity and permeability are unaffected.10.5 In general terms, it is possible to
say that the higher the strength of the hardened cement paste the lower its
permeability – a state of affairs to be expected because strength is a function of the
relative volume of gel in the space available to it. There is one exception to this
statement: drying the cement paste increases its permeability, probably because
shrinkage may rupture some of the gel between the capillaries and thus open new
passages to water.10.5
The difference between the permeability of hardened cement paste and of concrete
containing a paste of the same water/cement ratio should be appreciated as the
permeability of the aggregate itself affects the behaviour of the concrete (see Table
10.2). If the aggregate has a very low permeability, its presence reduces the effective
area over which flow can take place. Furthermore, because the flow path has to
circumvent the aggregate particles, the effective path becomes considerably longer so
that the effect of aggregate in reducing the permeability may be considerable. The
interface zone does not seem to contribute to flow. Generally, the influence of the
aggregate content in the mix is small and, because the aggregate particles are
enveloped by the cement paste, in fully compacted concrete it is the permeability of
the hardened cement paste that has the greatest effect on the permeability of the
concrete. This was referred to on p. 485.
The permeability of concrete under cryogenic conditions, e.g. to liquid nitrogen at
–196 °C, involves different mechanisms because ice reduces the flow and the
aggregate appears to have a substantial influence.10.50 Typical values of the intrinsic
permeability coefficient between 10–18 and 10–17 m2 have been reported.10.50
Permeability testing
Testing concrete for permeability has not been generally standardized10.123 so that the
values of the coefficient of permeability quoted in different publications may not be
comparable. In such tests as are used, the steady-state flow of water through concrete
due to a pressure differential is measured, and Darcy’s equation (see p. 486) is used to
calculate the coefficient of permeability, K.
The United States Bureau of Reclamation prescribes Procedure 4913-9210.43 in
which a water pressure of 2.76 MPa (400 psi) is used; this corresponds to a head of
water of 282 m. There exist also Canadian tests10.45,10.109 and a German test prescribed in
DIN 1048-1991.10.131 In these tests, the pressure under which water is forced to flow
through the concrete specimen is high, and this may alter the natural state of the
concrete; blocking of some pores by silting is also possible. Moreover, during the
progress of the test, hydration of the hitherto unhydrated cement can take place so that
the value of the calculated coefficient of permeability decreases with time.
The U.S. Bureau of Reclamation Procedure 4913-9210.43 provides for a correction
for the age of the specimen at test, as shown in Fig. 10.4. The Bureau of Reclamation
test is relevant to the behaviour of concrete in large dams. On the other hand, for usual
concrete structures, the flow of water under a high pressure is not representative of
service conditions.

537
Fig. 10.4. Correction for age in U.S. Bureau of Reclamation test for permeability
of concrete: the ordinate gives the permeability at any age as a percentage of
permeability at the age of 60 days10.43
It is important to note that the scatter of permeability test results made on similar
concrete at the same age, and using the same equipment, is large. Differences such as
between, say, 2 × 10–12 m/s and 6 × 10–12 m/s are not significant, so that reporting the
order of magnitude, or at the most the nearest 5 × 10–12 m/s, is adequate. Smaller
differences in the value of the coefficient of permeability are not significant and can
be misleading.
Water penetration test
There is a further problem with permeability testing, namely that, in good quality
concrete, there is no flow of water through the concrete. Water penetrates into the
concrete to a certain depth, and an expression has been developed by Valenta10.48 to
convert the depth of penetration into the coefficient of permeability, K (in metres per
second) equivalent to that used in Darcy’s law:

where e = depth of penetration of concrete in metres,

h = hydraulic head in metres,
t = time under pressure in seconds, and
v = the fraction of the volume of concrete occupied by pores.
The value of v represents discrete pores, such as air bubbles, which do not become
filled with water except under pressure, and can be calculated from the increase in the
mass of concrete during the test, bearing in mind that only the voids in the part of the
specimen penetrated by water should be considered. Typically, v lies10.47 between 0.02
and 0.06.

538
 The hydraulic head is applied by pressure which usually ranges between 0.1 and
0.7 MPa.10.21 The depth of penetration is found by observation of the split surface of the
test specimen (moist concrete being darker) after a given length of time. This is the
value of e in Valenta’s expression given above.
It is also possible to use the depth of penetration of water as a qualitative
assessment of concrete: a depth of less than 50 mm classifies the concrete as
‘impermeable’; a depth of less than 30 mm, as ‘impermeable under aggressive
conditions’.10.21
Air and vapour permeability
As mentioned earlier, the ease with which air, some gases, and water vapour can
penetrate into concrete is relevant to the durability of concrete under various
conditions of exposure. Distinction should be made between a situation when the
driving force is a pressure differential, on the one hand, and, on the other hand, a
situation when the pressure and temperature are the same on two sides of a concrete
specimen or member, but two different gases sweep the two sides. In the latter case,
the gases move through the concrete by diffusion whereas in the former we are
concerned with permeability.
Lawrence10.52 has reviewed the derivation and measurement of diffusivity of
concrete to gas, measured in square metres per second, and he has shown that, on a
log–log scale, diffusivity is linearly related to the intrinsic permeability of concrete,
measured in square metres. An example of this relation for oxygen is shown in Fig.
10.5. The relation can be exploited to establish the value of diffusivity from tests on
permeability, which are easier to perform.10.52

539
 Fig. 10.5. Relation between intrinsic permeability and diffusivity of concrete10.52
Because gases are compressible, the pressure, p0, at which the volume flow
rate, q (in m3/s), is measured, has to be taken into account, in addition to the inlet
pressure, p, and outlet pressure, pa; all pressures are absolute values in N/m2.
The intrinsic permeability coefficient, K, expressed in m2, is:10.96

where A = cross-sectional area of the specimen in m2,

L = its thickness in m, and
η = dynamic viscosity in N s/m2.
For oxygen at 20 °C, η = 20.2 × 10–6 N s/m2.
Theoretically, the intrinsic permeability coefficient of a given concrete should be
the same regardless of whether a gas or a liquid is used in the tests. However, gases
yield a higher value of the coefficient because of the phenomenon of gas slippage; this
means that, at the flow boundary, the gas has a finite velocity. The difference between

540
gas permeability and liquid permeability is larger at lower values of the intrinsic
permeability coefficient, the ratio of the former to the latter ranging from about 6 to
nearly 100.10.132
Air permeability is greatly affected by curing, especially in concretes of low and
moderate strengths.10.92 Figure 10.6 shows this effect for concrete cured for 28 days: (a)
in water, and (b) in air at a relative humidity of 65 per cent, and subsequently stored
for one year in air at 20 °C at a relative humidity of 65 per cent.

Fig. 10.6. Relation between oxygen permeability and compressive strength for
concretes cured for 28 days in water and in air at a relative humidity of 65 per
cent (based on ref. 10.92)
For the purposes of illustration,10.132 it can be mentioned that the order of magnitude
of intrinsic permeability (using gas) of concrete with a water/cement ratio of 0.33 is
10–18 m2.
The air permeability of concrete is strongly affected by its moisture content: a
change from near saturation to an oven-dried condition has been reported to increase
the gas permeability coefficient by nearly 2 orders of magnitude. For this reason, a
clearly defined condition of concrete should be used in all tests. From the standpoint
of the ease of testing, the oven-dried condition is preferred. However, this condition is
not representative of the concrete in service, and it is the permeability of concrete to

541
oxygen under actual conditions that is of importance in connection with corrosion of
reinforcing steel.
Conditioning a specimen in air at a constant relative humidity, even for as long as
28 days, does not necessarily result in a uniform moisture condition within the
concrete.10.59
The permeability of concrete to oxygen can be determined by a method developed
by Cembureau.10.53 However, there is no generally accepted test method.
Carbonation
Discussion of the behaviour of concrete is generally based on the assumption that the
ambient medium is air which does not react with hydrated cement paste. However, in
reality, air contains CO2 which, in the presence of moisture, reacts with hydrated
cement; the actual agent is carbonic acid because gaseous CO2 is not reactive.
The action of CO2 takes place even at small concentrations such as are present in
rural air, where the CO2 content is about 0.03 per cent by volume. In an unventilated
laboratory, the content may rise above 0.1 per cent; in large cities it is on average 0.3
per cent and, exceptionally, up to 1 per cent. An example of concrete exposed to a
very high concentration of CO2 is offered by the lining of vehicular tunnels. The rate
of carbonation of concrete increases with an increase in the concentration of CO2,
especially at high water/cement ratios,10.107 the transport of CO2 taking place through
the pore system in hardened cement paste.
Of the hydrates in the cement paste, the one which reacts with CO2 most readily is
Ca(OH)2, the product of the reaction being CaCO3, but other hydrates are also
decomposed, hydrated silica, alumina, and ferric oxide being
produced.10.7 Theoretically, such a complete decomposition of calcium compounds in
hydrated cement is chemically possible even at the low concentration of CO2 in the
normal atmosphere,10.101 but this is not a problem in practice. In concrete containing
Portland cement only, it is solely the carbonation of Ca(OH)2 that is of interest. When,
however, Ca(OH)2 becomes depleted, for instance by secondary reaction with
pozzolanic silica, the carbonation of calcium silicate hydrate, C-S-H, is also possible.
When this occurs, not only is more CaCO3 formed, but also the silica gel which is
concurrently formed has large pores, larger than 100 nm, which facilitates further
carbonation.10.67 The carbonation of C-S-H is discussed later in connection with
carbonation of concretes made with blended cements.
Effects of carbonation
Carbonation per se does not cause deterioration of concrete but it has important
effects. One of these is carbonation shrinkage, which was discussed on p. 444. With
respect to durability, the importance of carbonation lies in the fact that it reduces the
pH of the pore water in hardened Portland cement paste from between 12.6 to 13.5 to
a value of about 9. When all Ca(OH)2 has become carbonated, the value of pH is
reduced to 8.3.10.35 The significance of the lowering of the pH is as follows.
Steel embedded in hydrating cement paste rapidly forms a thin passivity layer of
oxide which strongly adheres to the underlying steel and gives it complete protection
from reaction with oxygen and water, that is from formation of rust or corrosion;
corrosion is discussed in Chapter 11. This state of the steel is known as passivation.
Maintenance of passivation is conditional on an adequately high pH of the pore water
in contact with the passivating layer. Thus, when the low pH front reaches the vicinity

542
of the surface of the reinforcing steel, the protective oxide film is removed and
corrosion can take place, provided oxygen and moisture necessary for the reactions of
corrosion are present. For this reason, it is important to know the depth of carbonation
and specifically whether the carbonation front has reached the surface of the
embedded steel. In fact, because of the presence of coarse aggregate, the ‘front’ does
not advance as a perfectly straight line. It might also be noted that, if cracks are
present, CO2 can ingress through them so that the ‘front’ advances locally from the
penetrated cracks. In many cases, corrosion can take place even when the full
carbonation front is still a few millimetres away from the surface of the steel if partial
carbonation has taken place.10.61
Rate of carbonation
Carbonation occurs progressively from the outside of concrete exposed to CO2, but
does so at a decreasing rate because CO2 has to diffuse through the pore system,
including the already carbonated surface zone of concrete. Such diffusion is a slow
process if the pores in hydrated cement paste are filled with water because the
diffusion of CO2 in water is 4 orders of magnitude slower than in air. On the other
hand, if there is insufficient water in the pores, CO2 remains in gaseous form and does
not react with the hydrated cement. It follows that the rate of carbonation depends on
the moisture content of the concrete, which varies with the distance from its surface.
Because of this variable situation, the rate of transport of CO2 to the advancing
carbonation front in the concrete cannot be readily determined from the diffusion
equation (see p. 487). The relation between diffusivity and intrinsic permeability
shown in Fig. 10.5 may possibly be exploited.
The highest rate of carbonation occurs at a relative humidity of between 50 and 70
per cent. This situation can be viewed against the background of a typical relative
humidity of 65 per cent in an ordinary laboratory; outdoors in southern England, the
average relative humidity is 86 per cent in winter and 73 per cent in summer.
Under steady hygrometric conditions, the depth of carbonation increases in
proportion to the square root of time, which is characteristic of sorption rather than
diffusion, but carbonation involves an interaction between CO2 and the pore system. It
is thus possible to express the depth of carbonation, D, in millimetres as

D = Kt0.5

where K = carbonation coefficient in mm/year0.5, and

t = time of exposure in years.
The values of K are often more than 3 or 4 mm/year0.5 for low-strength
concrete.10.58 Another way of giving a broad picture is to say that, in concrete with a
water/cement ratio of 0.60, a depth of carbonation of 15 mm would be reached after
15 years, but at a water/cement ratio of 0.45 only after 100 years. An example10.124 of
the progress of carbonation over a period of 16 years is shown in Fig. 10.7.

543
 Fig. 10.7. Progress of carbonation with time of exposure under different
conditions: (A) 20 °C and 65 per cent relative humidity; (B) outdoors, protected
by a roof; (C) horizontal surface outdoors in Germany. The values are averages
for concretes with water/cement ratios of 0.45, 0.60, and 0.80, wet-cured for 7
days (based on ref. 10.124)
The expression involving the square root of time is not applicable when the
exposure conditions are not steady. In particular, if the surface of the concrete is
exposed to a variable humidity, with periodic wetting, the rate of carbonation is
reduced because of a slowing down of the diffusion of CO2 through saturated pores in
the hardened cement paste. Conversely, sheltered parts of a structure undergo
carbonation at a faster rate than those exposed to rain, which significantly slows down

544
the progress of carbonation. In the interior of buildings, the rates of carbonation can
be high, but there are no ill consequences of this in so far as corrosion of embedded
steel is concerned (see p. 565) unless the carbonated concrete is subsequently wetted.
This can arise when water ingresses through the cladding of a building far enough to
reach the zone carbonated from inside.
The very considerable influence of the moisture content of the concrete upon
carbonation means that, even in a single building, made all of the same concrete, there
may be a considerable variation in the depth of carbonation at a given age: the walls
more exposed to rain will have a lower depth of carbonation; so will sloping surfaces
which can be washed down by rain; the same applies to walls which can be
thoroughly dried by strong insolation. Overall, the greatest depth of carbonation can
be 50 per cent more than the smallest depth.10.57
Small variations in temperature have little effect on carbonation but a high
temperature increases the rate of carbonation unless drying overshadows the
temperature effect.
The physico-chemical phenomena influencing the rate of carbonation are discussed
by Papadakis et al.10.56
Factors influencing carbonation
The fundamental factor controlling carbonation is the diffusivity of the hardened
cement paste. The diffusivity is a function of the pore system of the hardened cement
paste during the period when the diffusion of CO2 takes place. It follows that the type
of cement, the water/cement ratio, and the degree of hydration are relevant. All these
influence also the strength of the concrete containing any given hardened cement
paste. For this reason, it is often said that the rate of carbonation is simply a function
of the strength of concrete. This, while broadly true, is an inadequate simplification.
What makes the use of this approach worse is that the value of strength quoted is not
that applicable to the concrete in situ when it becomes exposed to CO2 but, usually,
the value of strength of test specimens cured in a standard manner, which is invariably
superior to curing in situ.
Alternatives to the use of strength as a parameter include expressing carbonation as
a function of the water/cement ratio or of the cement content, or of both of these.
There is no physical basis for considering the cement content; as far as the
water/cement ratio is concerned, such an approach is not superior to the use of
strength as a parameter. Indeed, neither strength nor water/cement ratio is informative
about the microstructure of the hardened cement paste in the surface zone of
concrete while the diffusion of CO2 is taking place. A factor which has a great
influence on the outer zone is the curing history of the concrete.
The effect of curing on carbonation of concrete is substantial. Figure 10.8 shows
the depth of carbonation of concretes with 28-day compressive strength (measured on
standard test cubes) between 30 and 60 MPa: (a) cured in water for 28 days, and (b)
cured in air at a relative humidity of 65 per cent; thereafter, all specimens were stored
for two years at 20 °C and a relative humidity of 65 per cent.10.92 The detrimental effect
of the absence of wet curing, which results in high porosity, is marked. Other research
workers10.133 have reported that increasing the period of wet curing from 1 day to 3 days
reduces the depth of carbonation by about 40 per cent.

545
Fig. 10.8. Relation between the depth of carbonation and compressive strength of
concrete after 2 years’ exposure in air at a relative humidity of 65 per cent
(based on ref. 10.92)
It should be noted, however, that outdoor exposure in many parts of the world
includes frequent or prolonged periods of high humidity so that hydration of cement
continues and, in effect, delayed natural curing of the surface zone takes place.
Nevertheless, generally, the effects of the absence of initial curing on carbonation
persist for many years in so far as it results in a microstructure of the hardened cement
paste in the outer zone of the concrete which facilitates the diffusion of CO2.
A general statement can be made to the effect that, in a situation conducive to
continuing carbonation, concrete with a strength lower than 30 MPa (4000 psi) is
highly likely to have undergone carbonation to a depth of at least 15 mm in a period
of several years.10.62
Despite the considerable variability in the rate of carbonation in different locations,
typical values reported by Parrott10.55 and shown in Table 10.4 are of interest. Clearly,

546
the values of Table 10.4 must not be treated as the norm. From Parrott’s data10.55 it is
possible to say that, for sheltered concrete outdoors in the United Kingdom or a
similar climate, in 90 per cent of cases, the depth of carbonation will not exceed the
values shown in Table 10.4. For the reasons given earlier, in some cases, the depth of
carbonation will exceed the 90 per cent upper bound; in others, it will be much lower.
Nevertheless, the typical values given in Tables 10.4 and 10.5, as well as the other
data presented in this chapter, should make it possible to ensure that the depth of
carbonation expected within the intended service life of the structure is smaller than
the cover to the reinforcement. Thus, the necessary depth of cover and the actual
quality of concrete are interdependent, in so far as protection of reinforcement is
concerned, so that at the design stage they should be chosen together. The topic of
cover is discussed on p. 575.

Table 10.4. Depth of Carbonation as a Function of Strength10.55

Table 10.5. Maximum Depth of Carbonation in Sheltered Concrete Outdoors in

the United Kingdom10.55

Carbonation of concrete containing blended cements

Because blended cements are widely used nowadays, it is important to know the
carbonation behaviour of concretes containing fly ash and ground granulated
blastfurnace slag. Numerous papers have been published reporting comparative
carbonation tests on concretes with and without these cementitious materials, but the
comparisons were made on varying bases. Such data do not lend themselves to useful
generalizations, and yet what is important when selecting a concrete mix is to assess
the carbonation characteristics of the specific proposed mix.
The starting point in this assessment is the knowledge of the microstructure and
other properties of the hardened cement paste resulting from the use of the various
cementitious materials in so far as these properties physically or chemically influence
carbonation. The relevant properties are discussed in Chapter 13, but in the present
context two observations should be made with respect to Class F fly ash. First of all,

547
the silica in the fly ash reacts with Ca(OH)2 resulting from the hydration of Portland
cement. In consequence, the blended cement leads to a lower Ca(OH)2 content in the
hardened cement paste so that a smaller amount of CO2 is required to remove all the
Ca(OH)2 by producing CaCO3. Bier10.67 has shown that the depth of carbonation is
greater when the amount of Ca(OH)2 present is lower. It follows that the presence of
fly ash results in a more rapid carbonation. There is, however, another effect of the
reaction between the pozzolanic silica and Ca(OH)2, namely it results in a denser
structure of the hardened cement paste so that its diffusivity is reduced and
carbonation is slowed down.
The question to ask is: which effect is dominant? One important factor is the
quality of curing. Good curing is necessary for the pozzolanic reactions to take place
(see p. 658), and yet tests involving only 1-day curing when concrete contains fly ash
have been made;10.55,10.66 such tests are destined to demonstrate high carbonation of
concrete with fly ash but they are predicated on bad concrete practice. The effects of
inadequate curing upon carbonation of concrete containing fly ash persist even in the
long term.10.63 On the other hand, concretes made with cement containing fly ash up to
30 per cent and with actual strengths above 35 MPa (5000 psi) have shown no
increase, or only a marginal increase, in carbonation when fly ash is included in the
mix.10.63,10.67
The use of ground granulated blastfurnace slag in the concrete mix entails an even
greater necessity for good curing. In consequence, poorly cured concrete containing
blastfurnace slag exhibits very high carbonation: depths of 10 to 20 mm after one
year’s exposure have been reported.10.64 High slag contents lead to a greater depth of
carbonation.10.64,10.65 However, when the blastfurnace slag content in the blended cement
is below 50 per cent and the concrete is exposed to CO2 at a concentration of 0.03 per
cent, there is only a marginal increase in carbonation.10.67
In view of the use of fillers in modern cements (see p. 88) it is useful to mention
that they have no effect on the microstructure of the hardened cement paste and,
therefore, they do not influence carbonation.10.60
Sulfate-resisting cement leads to a 50 per cent greater depth of carbonation than
Portland cement.10.108 For this reason, increased cover to reinforcement may be required
when sulfate-resisting cement is used. Carbonation of concrete containing regulated
set cement is also increased.10.137
Carbonation takes place also in high-alumina cement concrete but, as hydration of
that cement does not produce Ca(OH)2, it is calcium aluminate hydrates CAH10 and
C3AH6 that react with CO2. The end products are CaCO3 and alumina gel, which have
a lower strength than the hydrates. At the same strength as Portland cement concrete,
high-alumina cement concrete exhibits twice as high carbonation.10.124
The carbonation of hardened high-alumina cement paste may lead to the
depassivation of reinforcing steel, which in any case is in contact with pore water at a
pH lower than in the case of Portland cement, namely between 11.4 and 11.8. The rate
of carbonation of high-alumina-cement paste which has undergone conversion (see
p. 95) is much higher than prior to conversion.
Measurement of carbonation
Laboratory techniques which can be used to determine the depth of carbonation
include chemical analysis, X-ray diffraction, infra-red spectroscopy and
thermogravimetric analysis. A common and simple method to establish the extent of

548
carbonation is by treating a freshly broken surface of concrete with a solution of
phenolphthalein in diluted alcohol. The free Ca(OH)2 is coloured pink while the
carbonated portion is uncoloured; with progress of carbonation of the newly exposed
surface, the pink colouring gradually disappears. The procedure is prescribed by
RILEM.10.54 The test is easy to perform and is rapid but it should be remembered that
the pink colour indicates the presence of Ca(OH)2 but not necessarily a total absence
of carbonation. Indeed, the phenolphthalein test gives a measure of the pH (the colour
being pink above about 9.5) but does not distinguish between a low pH caused by
carbonation and by other acidic gases. As far as the risk of corrosion of reinforcement
is concerned, the cause of a low value of pH may not be important but care is required
in interpreting the observed colour pattern.
The phenolphthalein test cannot be used with high-alumina cement because, unlike
Portland cement, it does not contain free lime.
If breaking off a concrete sample is not practicable, dust samples of concrete can
be obtained by drilling to successively greater depths; the samples are then subjected
to the phenolphthalein test. Care is required because, if free lime from uncarbonated
concrete contaminates a sample, the entire sample will turn pink, giving the
impression of an absence of carbonation.
Under some circumstance, measurement of the carbonation front from crack
surfaces could be used to show that the crack is old; when cracks of known age exist,
comparisons could indicate the age of the given crack.10.140
To determine how rapidly a given concrete is likely to undergo carbonation,
accelerated testing can be used. This consists of exposing a concrete specimen to
concentrated CO2 of cs per cent. The depth of carbonation after a certain period of
exposure, tt, can then be transformed into an estimate of the length of time, ts, for the
same depth to be reached at the service concentration of CO2 of cs per cent, on the
basis that the time is inversely proportional to the concentration of CO2:

tt:ts = cs:100.

In the Swiss method,10.125 now discontinued, test concentration of CO2 of 100 per
cent is used; more often the concentration is 4 or 5 per cent.10.36 For carbonation to
proceed, the relative humidity should be 60 to 70 per cent.
Considerable care is required in interpreting accelerated tests, not only because in
situ carbonation is greatly influenced by the exact exposure conditions, especially
with respect to wetting by rain and drying by sun and wind, but also because a high
concentration of CO2 distorts the phenomena involved. For example, Bier,10.67 using a
concentration of CO2 of 2 per cent, found that the depth of carbonation of well-cured
concrete containing fly ash or blastfurnace slag is greater by a factor of at least 2 than
when Portland cement only is present. There was no such increase in the depth of
carbonation where the CO2 concentration was 0.03 per cent and the content of fly ash
was below 30 per cent or of blastfurnace slag below 50 per cent. A likely explanation
of this difference in behaviour is that, at a high concentration of CO2, the carbonation
of Ca(OH)2 was followed by the carbonation of C-S-H.
Extensive carbonation of C-S-H in concrete in service has been reported by
Kobayashi et al.,10.110 but information on the type of cement used is not available.
Further aspects of carbonation

549
Carbonation can have some positive consequences. Because CaCO3 occupies a greater
volume than Ca(OH)2 which it replaces, the porosity of carbonated concrete is
reduced. Also, water released by Ca(OH)2 on carbonation may aid the hydration of
hitherto unhydrated cement. These changes are beneficial and they result in increased
surface hardness, increased strength at the surface,10.104 reduced surface
permeability,10.102 reduced moisture movement,10.103 and increased resistance to those
forms of attack which are controlled by permeability. On the other hand, carbonation
accelerates chloride-induced corrosion of reinforcement (see p. 572).
Unlike Portland cement concrete, with supersulfated cement there is a loss of
strength on carbonation but, because this applies to the surface zone of the concrete
only, the loss is not structurally significant.
Because carbonation affects the porosity and also the pore size distribution
(causing a decrease in the volume of pores, especially of the smaller ones) of the outer
zone of concrete, the penetration of paint into concrete will vary. In consequence, the
bond of paint and the colouring are affected by carbonation.10.100 Because the latter
depends on the relative humidity of air and on age, it is easy to see how differences in
colouring and quality of painting can readily arise.
Sakuta et al.10.138 have proposed the use of additives that absorb carbon dioxide
which has entered the concrete, thereby preventing carbonation.
Acid attack on concrete
Concrete is generally well resistant to chemical attack, provided an appropriate mix is
used and the concrete is properly compacted. There are, however, some exceptions.
First of all, concrete containing Portland cement, being highly alkaline, is not
resistant to attack by strong acids or compounds which may convert to acids.
Consequently, unless protected, concrete should not be used when this form of attack
may occur.
Generally speaking, chemical attack of concrete occurs by way of decomposition
of the products of hydration and formation of new compounds which, if soluble, may
be leached out and, if not soluble, may be disruptive in situ. The attacking compounds
must be in solution. The most vulnerable cement hydrate is Ca(OH)2, but C-S-H can
also be attacked. Calcareous aggregates are also vulnerable.
Of the common forms of attack, that by CO2 was considered in the preceding
section, whereas the attack by sulfates and the action of sea water will be discussed
later in the present chapter. Comprehensive lists of substances which attack concrete
to varying degrees can be found in ACI 515.1R (Revised 1985),10.93 ACI 201.2R-
92,10.42 and in a book by Biczok.10.71 A limited composite extract is given in Table 10.6.
Additionally, specific mention of some aggressive substances is made below.

Table 10.6. A List of Some Substances which Cause Severe Chemical Attack of
Concrete

550
 Concrete can be attacked by liquids with a pH value below 6.510.26 but the attack is
severe only at a pH below 5.5; below 4.5, the attack is very severe. A concentration of
CO2 of 30 to 60 ppm results in a severe attack, and above 60 ppm results in a very
severe attack.
The attack progresses at a rate approximately proportional to the square root of
time because the attacking substance has to travel though the residual layer of the
low-solubility products of reaction which remain after Ca(OH)2 has been dissolved.
Thus it is not only pH but also the ability of aggressive ions to be transported that
influence the progress of the attack.10.26 Also, the rate of attack decreases when
aggregate has become exposed because the vulnerable surface is smaller and the
attacking medium has to travel around the particles of aggregate.10.26
Concrete is also attacked by water containing free CO2, such as moorland water or
mineral waters, which may also contain hydrogen sulfide. Not all CO2 is aggressive
because some of it is required to form and stabilize calcium bicarbonate in the
solution. Flowing pure water, formed by melting ice or by condensation (for instance,
in a desalination plant) and containing little CO2, also dissolves Ca(OH)2, thus causing
surface erosion. Peaty water with CO2 is particularly aggressive; it can have a pH
value as low as 4.4.10.31 This type of attack may be of importance in conduits in
mountain regions, not only from the standpoint of durability but also because the
leaching out of hydrated cement leaves behind protruding aggregate and increases the
roughness of the pipe. To avoid this, the use of calcareous, rather than siliceous,
aggregate is advantageous10.10 because both the aggregate and the cement paste are
eroded.
Acid rain, which consists mainly of sulfuric acid and nitric acid and has a pH value
between 4.0 and 4.5, may cause surface weathering of exposed concrete.10.70
Although domestic sewage by itself is alkaline and does not attack concrete, severe
damage of sewers has been observed in many cases, especially at fairly high
temperatures,10.10 when sulfur compounds become reduced by anaerobic bacteria to H2S.
This is not a destructive agent in itself, but it is dissolved in moisture films on the
exposed surface of the concrete and undergoes oxidation by aerobic bacteria, finally
producing sulfuric acid. The attack occurs, therefore, above the level of flow of the
sewage. The hardened cement paste is gradually dissolved, and progressive

551
deterioration of concrete takes place.10.27 A somewhat similar form of attack can occur
in offshore oil storage tanks.10.134
Sulfuric acid is particularly aggressive because, in addition to the sulfate attack of
the aluminate phase, acid attack on Ca(OH)2 and C-S-H takes place. Reduction in the
cement content of the concrete is therefore beneficial,10.78 provided, of course, that the
density of the concrete is unimpaired.
Concrete is generally resistant to microbiological attack because its high pH does
not encourage such action; nevertheless, under certain, fortunately rare, tropical
conditions, some algae, fungi and bacteria can use atmospheric nitrogen to form nitric
acid which attacks concrete.10.73
Lubricating oils and hydraulic fluid, sometimes spilt on concrete aprons at airports,
break down when heated by exhaust gases and react with Ca(OH)2, thus causing
leaching.10.69
Various physical and chemical tests on the resistance of concrete to acids have
been developed,10.7 but there are no standard procedures. It is essential that tests are
performed under realistic conditions because, when a concentrated acid is used, all
cements dissolve and no assessment of their relative quality is possible. For this
reason, care is required in interpreting the results of accelerated tests.
The need for testing under specific conditions arises from the fact that pH alone is
not an adequate indicator of the potential attack: the presence of CO2 in relation to the
hardness of water also influences the situation. An increase in the rate of flow of the
attacking medium, and in its temperature and pressure, all increase the attack.
Use of blended cements which include ground granulated blastfurnace slag,
pozzolanas, and especially silica fume, is beneficial in reducing the ingress of
aggressive substances. Pozzolanic action also fixes Ca(OH)2, which is usually the
most vulnerable product of hydration of cement in so far as acid attack is concerned.
However, the performance of concrete depends more on its quality than on the type of
cement used. The resistance of concrete to chemical attack is increased by allowing it
to dry out before exposure, but following proper curing. A thin layer of calcium
carbonate (produced by the action of CO2 on lime) then forms, blocking the pores and
reducing the permeability of the surface zone. It follows that precast concrete is
generally less vulnerable to attack than concrete cast in situ. Good protection of
concrete from acid attack is obtained by subjecting precast concrete in a vacuum to
the action of silicon tetrafluoride gas.10.11 This gas reacts with lime:

2Ca(OH)2 + SiF4 → 2CaF2 + Si(OH)4.

Ca(OH)2 can also be fixed by treatment with diluted water-glass (sodium silicate).
Calcium silicates are then formed, filling the pores. Treatment with magnesium
fluorosilicate is also possible. The pores become filled and the resistance of the
concrete to acid is also slightly increased, probably due to the formation of colloidal
silicofluoric gel. There exist numerous surface treatment methods10.93 but this subject is
outside the scope of the present book.
Sulfate attack on concrete
Solid salts do not attack concrete but, when present in solution, they can react with
hydrated cement paste. Particularly common are sulfates of sodium, potassium,
magnesium, and calcium which occur in soil or in groundwater. Because the solubility
of calcium sulfate is low, groundwaters with a high sulfate content contain the other

552
sulfates as well as calcium sulfate. The significance of this lies in the fact that those
other sulfates react with the various products of hydration of cement and not only with
Ca(OH)2.
Sulfates in groundwater are usually of natural origin but can also come from
fertilizers or from industrial effluents. These sometimes contain ammonium sulfate,
which attacks hydrated cement paste10.83 by producing gypsum.10.95 Soil in some disused
industrial sites, particularly gas works, may contain sulfates and often also other
aggressive substances. Sulfides can oxidize to sulfates under some conditions, e.g.
under compressed air used in excavation.
The reactions of the various sulfates with hardened cement paste are as follows.
Sodium sulfate attacks Ca(OH)2:

Ca(OH)2 + Na2SO4.10H2O → CaSO4.2H2O + 2NaOH + 8H2O.

This is an acid-type attack. In flowing water, Ca(OH)2 can be completely leached out
but if NaOH accumulates, equilibrium is reached, only a part of the SO3 being
deposited as gypsum.
The reaction with calcium aluminate hydrate can be formulated as follows:10.7

2(3CaO.Al2O3.12H2O) + 3(Na2SO4.10H2O) → 3CaO.Al2O3.3CaSO4.32H2O +

2Al(OH)3 + 6NaOH + 17H2O.

Calcium sulfate attacks only calcium aluminate hydrate, forming calcium

sulfoaluminate (3CaO.Al2O3.3CaSO4.32H2O), known as ettringite. The number of
molecules of water may be 32 or 31, depending upon the ambient vapour pressure.10.74
On the other hand, magnesium sulfate attacks calcium silicate hydrates as well as
Ca(OH)2 and calcium aluminate hydrate. The pattern of reaction is:

3CaO.2SiO2.aq + 3MgSO4.7H2O → 3CaSO4.2H2O + 3Mg(OH)2 + 2SiO2.aq.

+ xH2O.

Because of the very low solubility of Mg(OH)2, this reaction proceeds to completion
so that, under certain conditions, the attack by magnesium sulfate is more severe than
by other sulfates. A further reaction between Mg(OH)2 and silica gel is possible and
may also cause deterioration.10.23 The critical consequence of the attack by magnesium
sulfate is the destruction of C-S-H.
Thaumasite form of sulfate attack
This type of attack occurs in concrete buried in the ground. Thaumasite problems
have been encountered in a number of bridge supports in the UK but they are not very
common. At temperatures below about 15 °C (59 °F) in the presence of sulfate,
carbonate and water, C-S-H can convert to thaumasite, which is a non-binder, with a
composition of CaSiO3.CaCO3.CaSO4.15H2O.10.139 The carbonate may be present in the
aggregate (limestone or dolomite) or as bicarbonate in the groundwater. Mixes
containing ggbs offer resistance to thaumasite attack.
Delayed ettringite formation
This phenomenon (known as DEF) came into prominence in the 1990s and it attracted
much academic research; the interest has abated somewhat since then.

553
 The formation of ettringite in the expansive cement Type K was discussed on
p. 448. This is an early-age controlled expansion. However, the formation of ettringite
in mature concrete tends to be disruptive and harmful, and is a form of sulfate attack,
resulting in the compound 3CaO.Al2O3.3CaSO4.32H2O.
High temperature during hydration can be the result of applied heat or be due to
heat generation in a large pour when natural loss of heat is inadequate. If the
temperature in the interior of concrete reaches 70° to 80 °C, a slow formation
of ettringite can lead to expansion and cracking. For the harmful effects to take place
after return to room temperature, the concrete has to be wet or moist either
permanently or intermittently.10.143 These harmful effects are a loss of strength, a
decrease in the modulus of elasticity, and sometimes cracking.
Occasionally, there is a problem of distinguishing DEF from alkali-silica reaction.
This was the issue in a court case concerning prestressed steam-cured railway sleepers
(ties). One reason for confusion is that ettringite may be so fine as to look like alkali-
silica gel.10.144
DEF is often avoidable by a selection of appropriate blended cement that does not
lead to an excessive temperature rise.
Mechanisms of attack
The mechanism of the delayed ettringite expansion is still debated, there being two
principal schools of thought.
Mather10.81 and many others are of the opinion that the reaction between calcium
sulfate and C3A is topochemical, that is, it is a solid-state reaction, not involving
solution and re-precipitation which would allow movement of the newly formed
product away from the original location. Such movement would not result in the
development of pressure. If the product of the topochemical reaction occupies a larger
volume than the volume of the two original compounds, then expansive and
disruptive forces are created. In the case of the reaction between calcium sulfate and
Ca(OH)2, there is no overall increase in volume10.74 but, because of the differences in
the solubility of C3A and gypsum, oriented, acicular ettringite is formed at the surface
of the C3A. Thus, there is a local increase in volume and, at the same time, an increase
in porosity elsewhere.10.75
The second school of thought whose chief protagonist is Mehta10.83 attributes the
development of expansive forces to the swelling pressure induced by the adsorption of
water by the originally colloidal ettringite which precipitated in the solution in the
presence of lime. Thus, the formation of ettringite per se is thought to be the cause of
expansion. However, Odler and Glasser10.75 point out that an uptake of water from the
environment is not a necessary condition for expansion to take place. Nevertheless,
expansion increases significantly under wet conditions10.75 so that it is likely that both
the mechanisms of expansion discussed above are involved at different stages.10.82 It
should be added that the concept of expansive forces induced by crystallization per se,
advanced by some researchers, seems erroneous.
Ettringite can also form from the reaction between sulfate and C4AF, but this
ettringite is nearly amorphous and no damaging expansion has been
reported.10.75 Nevertheless, ASTM C 150-09 prescribes a limit on the combined content
of C3A and C4AF (see p. 76), when sulfate resistance is required.
The consequences of sulfate attack include not only disruptive expansion and
cracking, but also loss of strength of concrete due to the loss of cohesion in the

554
hydrated cement paste and of adhesion between it and the aggregate particles.
Concrete attacked by sulfates has a characteristic whitish appearance. The damage
usually starts at edges and corners and is followed by progressive cracking and
spalling which reduce the concrete to a friable or even soft state.
The attack occurs only when the concentration of the sulfates exceeds a certain
threshold. Above that, the rate of sulfate attack increases with an increase in the
strength of the solution, but beyond a concentration of about 0.5 per cent of MgSO4 or
1 per cent of Na2SO4 the rate of increase in the intensity of the attack becomes
smaller.10.7 A saturated solution of MgSO4 leads to serious deterioration of concrete,
although with a low water/cement ratio this takes place only after 2 to 3 years.10.13 BS
EN 206-1 : 2000 expresses sulfate as SO3 while ACI uses SO4; multiplying the former
by 1.2 converts it into the latter. Water-soluble sulfates, and not acid-soluble, are
considered The classification of the severity of exposure recommended by ACI 318-
0810.42 and BS EN 206-1 : 2000 is given in Table 10.7. Because extraction of sulfates
from soil depends on the compaction of the soil, and on the water-soil extraction ratio,
the measurement of sulfates in groundwater is more reliable. The class boundaries are,
in a sense, arbitrary because they have not been calibrated by measurement of
recorded incidence of damage to concrete caused by sulfate attack. Moreover, the
actual conditions of exposure may vary during the lifetime of a structure owing to a
variation in groundwater flow or the drainage pattern.

Table 10.7. Classification of Severity of Sulfate Environment

The classification of the severity of exposure recommended by ACI 201.2R-

92 is given in Table 10.7. The approach of BS 8110-1 : 1985 (superseded by
10.42

Eurocode 2 : 2008) is somewhat more elaborate in that there are more subdivisions
corresponding to the “severe” exposure condition of ACI 201.2R-92.
It should be noted that, under certain conditions, the sulfate concentration in water
can be greatly increased by evaporation. This is the case with sea water splash on
horizontal surfaces and on the surface of cooling towers.10.79
In addition to the concentration of the sulfate, the speed with which concrete is
attacked depends also on the rate at which the sulfate removed by the reaction with
cement can be replenished. Thus, in estimating the danger of sulfate attack, the
movement of groundwater has to be known. When concrete is exposed to the pressure

555
of sulfate-bearing water on one side, the rate of attack will be highest. Likewise,
alternating saturation and drying leads to rapid deterioration. On the other hand, when
the concrete is completely buried, without a channel for the groundwater, conditions
will be much less severe.
Factors mitigating the attack
The purpose of the classification of the severity of sulfate exposure shown in Table
10.7 is to suggest preventive measures. Two approaches can be used. The first one is
to minimize the C3A content in the cement, that is, to use sulfate-resisting cement; this
is discussed on p. 76. The second approach is to reduce the quantity of Ca(OH)2 in
hydrated cement paste by the use of blended cements containing blastfurnace slag or
pozzolana. The effect of pozzolana is two-fold. First, it reacts with Ca(OH)2 so that
Ca(OH)2 is no longer available for reaction with sulfates. Second, compared with
Portland cement only, the same content of blended cement per cubic metre of concrete
results in less Ca(OH)2. These measures are helpful, but even more important is the
prevention of ingress of sulfates into the concrete: this is achieved by making the
concrete as dense as possible and with as low a permeability as possible. This must
never be forgotten: for instance, the use of lean concrete in the haunching or bedding
of sewers produces vulnerable parts of a possibly otherwise durable construction.
As far as the choice of cement is concerned, ACI 201.2R-9210.42 recommends, for
moderate exposure, the use of Type II cement, or of blended cement with blastfurnace
slag or pozzolana. For severe exposure, sulfate-resisting cement is the preferred
choice; for very severe exposure, a blend of sulfate-resisting cement and pozzolana
(between 25 and 40 per cent by mass of total cementitious material) or blastfurnace
slag (not less than 70 per cent by mass) proven to improve sulfate resistance, is
required.10.135 The relevant property of the blast-furnace slag is its alumina
content;10.80 advice on this is given in ASTM C 989-09a and in ref. 10.135. It should
also be noted that not all pozzolanas are beneficial: a low calcium oxide content is
desirable;10.77 specifically, Class C fly ash decreases the sulfate resistance of
concrete.10.76
The reason why sulfate-resisting cement alone is inadequate under severe
conditions is that not only calcium sulfate but also other sulfates are present.
Therefore, although sulfate-resisting cement does not contain enough C3A for the
formation of expansive ettringite, the Ca(OH)2 present and possibly also C-S-H are
vulnerable to the acid-type attack of the sulfates.
The recommendations of ACI 201.2R-9210.42 reflect the beneficial effect on sulfate
resistance of pozzolanas and ground granulated blastfurnace slag used with Portland
cement. Pozzolanas have also to be used with regulated-set cement, which alone
shows a poor resistance to sulfates. However, partial replacement (20 per cent) of this
cement by pozzolanas reduces the early strength of concrete10.24 so that the practicality
of use of regulated-set cement under conditions of sulfate attack is questionable.
Silica fume incorporated in concrete is beneficial with respect to permeability, but
tests on hardened cement paste indicate that the effect of silica fume in various sulfate
environments is not clear.10.126
Supersulfated cement offers very high resistance to sulfates, especially if its
Portland cement component is of the sulfate-resisting variety.
High-pressure steam curing improves the resistance of concrete to sulfate attack.
This applies to concretes made both with sulfate-resisting and ordinary Portland

556
cements because the improvement is due to the change of C3AH6 into a less reactive
phase, and also to the removal of Ca(OH)2 by the reaction with silica.
It is worth noting that, because of changes in solubility with temperature,
expansion due to the formation of ettringite is very low at temperatures above 30 °C
(86 °F).10.127
Low permeability of concrete, as mentioned earlier in this chapter, is the
consequence of an appropriate microstructure of the hardened cement paste. In order
to achieve this, the mix proportions need to be specified. There are three possible
approaches, one or more of these being used by various codes: specifying a maximum
water/cement ratio, specifying a minimum strength, and specifying a minimum
cement content. The same choice applies when a low permeability of concrete is
sought in cases of protection from other forms of attack.
The concept of ensuring protection from sulfate attack by specifying a minimum
cement content has no scientific basis. As Mather10.25 points out, for instance, with 356
kg of ordinary Portland cement per cubic metre of concrete (600 lb/yd3) it is possible
to obtain concretes ranging in cylinder strength from 14 MPa (2000 psi) to 41 MPa
(6000 psi) depending on the water/cement ratio and on slump. The durability of these
concretes will clearly vary enormously.
The use of strength for specifying purposes is convenient but strength only reflects
the water/cement ratio; it is this that is relevant to density and permeability, as
discussed on p. 272. However, specifying the water/cement ratio regardless of the
nature of the cement used is inadequate: reference to the influence of the various
blended cements on sulfate resistance was made earlier in this section.
Tests on sulfate resistance
The resistance of concrete to sulfate attack can be tested in the laboratory by storing
specimens in a solution of sodium or magnesium sulfate, or in a mixture of the two.
Alternate wetting and drying accelerates the damage due to the crystallization of salts
in the pores of the concrete. The effects of exposure can be estimated by the loss in
strength of the specimen, by changes in its dynamic modulus of elasticity, by its
expansion, by its loss of mass, or can even be assessed visually.
Figure 10.9 shows the change in the dynamic modulus of 1:3 mortar immersed
(after 78 days’ moist curing) in a 5 per cent solution of different sulfates.10.9 The test
method of ASTM C 1012-09 uses immersion of well-hydrated mortar in a sulfate
solution, and considers excessive expansion as a criterion of failure under sulfate
attack. This test can be used to assess the effects of using various cementitious
materials in the mix. As, however, it is mortar and not concrete that is tested, some
physical effects of materials such as silica fume or fillers are not reflected in the test.
A further drawback of the test is that it is slow, several months sometimes being
required before failure is recorded or its absence can be inferred.

557
 Fig. 10.9. Effect of immersion in a 5 per cent sulfate solution on the dynamic
modulus of elasticity of 1:3 mortars made with ordinary Portland and
supersulfated cements10.9
As an alternative to immersion in a sulfate solution, ASTM C 452-06 prescribes a
method in which a certain amount of gypsum is included in the original mortar mix.
This speeds up the reaction with C3A but the method is not appropriate for use with
blended cements, in which some cementitious material is still unhydrated at the stage
of coming into contact with the sulfates. This is so because, in the test of ASTM C
452-06, the criterion of sulfate resistance is the expansion at the age of 14 days.
It may be relevant to mention one more test, namely, ASTM C 1038-04, which
determines the expansion of mortar made from Portland cement of which sulfate is an
integral part. Thus, the test identifies excessive sulfate content of a Portland cement
rather than attack by external sulfates.
All the ASTM tests are performed on mortars of prescribed proportions and, in
consequence, are more sensitive to the chemical resistance of cement than to the
physical structure of the hardened cement paste in the actual concrete.
Efflorescence
Leaching of lime compounds, mentioned earlier, may under some circumstances lead
to the formation of salt deposits on the surface of the concrete, known as
efflorescence. This is found, for instance, when water percolates through poorly
compacted concrete or through cracks or along badly made joints, and when
evaporation can take place at the surface of the concrete. Calcium carbonate formed
by the reaction of Ca(OH)2 with CO2 is left behind in the form of a white deposit.
Calcium sulfate deposits are encountered as well.

558
 Efflorescence is more likely to occur in concrete which is porous near the surface.
Thus, the type of formwork may play a role in addition to the degree of compaction
and to the water/cement ratio.10.28 The occurrence of efflorescence is greater when cool,
wet weather is followed by a dry and hot spell; in this sequence, there is little initial
carbonation, lime is dissolved by the surface moisture, and Ca(OH)2 is finally drawn
to the surface.10.28
Efflorescence can also be caused by the use of unwashed seashore aggregate. The
salt coating on the surface of the aggregate particles may, in due course, lead to a
white deposit on the surface of the concrete. Gypsum and alkalis in the aggregate
have a similar effect. Transport of salts from the ground through porous concrete to a
drying surface can also result in efflorescence.
Apart from the leaching aspect, efflorescence is of importance only in so far as it
mars the appearance of concrete.
Early efflorescence can be removed with a brush and water. Heavy deposits may
require acid treatment of the surface of the concrete. Such treatment can also be used
to remove laitance on architectural concrete and to restore the roughness of floor
surfaces.10.29 The acid used is HCl diluted from its concentrated form in a ratio of 1:20
or 1:10. Typically, the thickness of the layer of acid (applied by a sponge) would be
0.5 mm (0.02 in.), and the quantity of a 1:10 solution of the acid used would be 200
g/m2, and the depth of concrete removed about 0.01 mm (0.0004 in.). The action of
the acid stops when it has been used up by the reaction with lime, but the concrete
should be washed in order to remove the salts which have been formed.10.29
Because lime is removed by the acid, the surface of the concrete becomes darker.
For this reason, as well as to avoid local ‘excesses’, the acid must be applied
uniformly in terms of concentration, quantity, and duration of action. Acid treatment
is a very delicate operation and trying it out on concrete samples is essential.
Another surface blemish is the appearance of dark stains of irregular shape which
are visible depending on the direction of light. Their origin is totally different from
efflorescence: they are compacts of cement paste, almost without pores. These can be
caused by the aggregation of coarse particles of cement which have hydrated only
little in locations where the water/cement ratio is very low. It is the lack of hydration
and of production of lime that leads to the dark colour. Such a segregation of coarse
particles of cement can be caused by a filtering action of leaky formwork or of
aggregate particles. With time, hydration may take place and the dark colour may
disappear.10.30
Effects of sea water on concrete
Concrete exposed to sea water can be subjected to various chemical and physical
actions. These include chemical attack, chloride-induced corrosion of steel
reinforcement, freeze–thaw attack, salt weathering, and abrasion by sand in
suspension and by ice. The presence and intensity of these various forms of attack
depend on the location of the concrete with respect to the sea level. These forms of
attack will be considered later in this chapter and in Chapter 11, starting with
chemical attack, which is the subject matter of this section.
Chemical action of sea water on concrete arises from the fact that sea water
contains a number of dissolved salts. The total salinity is typically 3.5 per cent.
Specific values are: 0.7 per cent in the Baltic Sea, 3.3 per cent in the North Sea, 3.6
per cent in the Atlantic and Indian Oceans, 3.9 per cent in the Mediterranean Sea, 4.0

559
per cent in the Red Sea, and 4.3 per cent in the Persian–Arabian Gulf. In all the seas,
the ratio of the individual salts is very nearly constant; for example, in the Atlantic
Ocean, the ion concentration (in per cent) is as follows: chloride 2.00, sulfate 0.28,
sodium 1.11, magnesium 0.14, calcium 0.05, and potassium 0.04. Sea water contains
also some dissolved CO2. Shallow coastal areas in hot climates, where evaporation is
high, can be very salty. The Dead Sea is the extreme case: its salinity is 31.5 per cent,
that is, nearly 9 times that of the oceans, but the sulfate concentration is lower than in
the oceans.10.91
The pH of sea water varies between 7.5 and 8.4, the average value in equilibrium
with atmospheric CO2 being 8.2.10.79 Ingress of sea water into concrete per se does not
significantly lower the pH of pore water in the hardened cement paste: the lowest
value reported is 12.0.10.86
The presence of a large quantity of sulfates in sea water could lead to the
expectation of sulfate attack. Indeed, the reaction between sulfate ions and both C3A
and C-S-H takes place, resulting in the formation of ettringite, but this is not
associated with deleterious expansion because ettringite, as well as gypsum, are
soluble in the presence of chlorides and can be leached out by the sea water.10.7 It
follows that the use of sulfate-resisting cement in concrete exposed to the sea is not
essential, but a limit on C3A of 8 per cent when the SO3 content is less than 3 per cent,
is recommended; cements with a C3A content up to 10 per cent can be used, provided
the SO3 content does not exceed 2.5 per cent.10.90 It seems that it is the excess of
SO3 that leads to a delayed expansion of concrete. The same tests10.90 confirmed that
C4AF also leads to the formation of ettringite so that the requirement of the earlier
version of ASTM C 150-09 to the effect that the content of 2C3A + C4AF be less than
25 per cent of the clinker for sulfate-resisting cement should be observed.
The preceding comments and requirements apply to concrete permanently
immersed in water, which represents relatively protected exposure
conditions10.88 because a steady state of saturation and salt concentration is reached so
that the diffusion of ions is greatly reduced. Alternating wetting and drying represents
much more severe conditions because a build-up of salts within the concrete can
occur in consequence of the ingress of sea water, followed by evaporation of pure
water, with the salts left behind. As the most damaging effect of sea water on concrete
structures arises from the action of chlorides on the steel reinforcement, the build-up
of salts will be discussed in Chapter 11 in the section on chloride attack.
The chemical action of sea water on concrete is as follows. The magnesium ion
present in the sea water substitutes for the calcium ion:

MgSO4 + Ca(OH)2 → CaSO4 + Mg(OH)2.

The resulting Mg(OH)2, known as brucite, precipitates in the pores at the surface of
the concrete, thus forming a protective surface layer which impedes further reaction.
Some precipitated CaCO3 in the form of aragonite, arising from the reaction of
Ca(OH)2 with CO2, may also be present. The precipitated deposits, typically 20 to
50 μm thick, form rapidly;10.84 they have been observed in a number of fully submerged
sea structures. The blocking nature of brucite makes its formation self-limiting.
However, if abrasion can remove the surface deposit, then the reaction by the
magnesium ion freely available in the sea water continues.
This situation is an example of the synergistic action of the different modes of
attack by the sea: wave action enhances chemical attack, and chemical attack by way

560
of formation and crystallization of salts makes the concrete more vulnerable to
erosion by wave action and to abrasion by sand suspended in sea water.
Salt weathering
When concrete is repeatedly wetted by sea water, with alternating periods of drying
during which pure water evaporates, some of the salts dissolved in sea water are left
behind in the form of crystals, mainly sulfates. These crystals re-hydrate and grow
upon subsequent wetting, and thereby exert an expansive force on the surrounding
hardened cement paste. Such progressive surface weathering, known as salt
weathering, occurs in particular when the temperature is high and insolation is strong
so that drying occurs rapidly in the pores over some depth from the surface. Thus,
intermittently wetted surfaces are vulnerable; these are surfaces of the concrete in the
tidal zone and in the splash zone. Horizontal or inclined surfaces are particularly
prone to salt weathering, and so are surfaces wetted repeatedly but not at short
intervals so that thorough drying can take place. Salt water can also rise by sorption,
that is, by capillary action; evaporation of pure water from the surface leaves behind
salt crystals which, when re-wetted, can cause disruption.
Salt weathering can occur not only in consequence of direct spray by sea water, but
also when air-borne salt which has been deposited on the surface of the concrete
becomes dissolved by dew, and this is followed by evaporation. Such behaviour has
been observed in desert areas where the large temperature drop in the small hours of
the night reduces the relative humidity of the air to the point where condensation in
the form of dew occurs. Salt weathering can extend to a depth of several millimetres:
hardened cement paste and the embedded fine aggregate particles are removed,
leaving behind protruding coarse aggregate particles. With time, these particles can
become loosened, thereby exposing more hardened cement paste which, in turn,
becomes liable to salt weathering. The process is, in essence, similar to the salt
weathering of porous rocks. Even when sodium sulfate is involved, the damage
mechanism is physical so that it does not represent sulfate attack.
It should be added that, unless the aggregate is dense and has very low absorption,
the aggregate itself is liable to damage. Clearly, such aggregate should not be used in
concrete exposed to conditions conducive to salt weathering so that the choice of
suitable aggregate is of considerable importance.10.85 Because the attack of concrete by
salt weathering is physical in nature, the type of cement used is of little
importance per se but, to ensure low permeability of the surface zone of concrete, the
choice of the concrete mix is critical.
Salt weathering can also result from the use of de-icing salts on concrete surfaces
in cold climates. This is known as salt scaling; this topic is discussed in Chapter 11.
A peculiar form of marine attack of concrete in very warm sea water has been cited
by Bijen.10.129 When limestone aggregate is present, a genus of oysters and also a genus
of sponges devour lime and produce holes up to 10 mm in diameter and 150 mm deep.
The rate of attack is up to 10 mm per annum.
Selection of concrete for exposure to sea water
The preceding discussion of the various modes of attack by sea water has emphasized
the importance of the low permeability of the exposed concrete. This can be achieved
by the use of a low water/cement ratio, an appropriate choice of cementitious
materials, good compaction, and absence of cracking due to shrinkage, thermal effects,
or stresses in service. It is important for the concrete to be well cured prior to

561
exposure to the sea water. The assumption that sea water also provides curing is
erroneus (see p. 574) unless the concrete, once immersed in sea water, remains
permanently submerged. Tests on mortar have led to a recommendation of a
minimum period of seven days of curing in fresh water, regardless of the type of
cement used.10.89
Reference to the choice of cement was made on p. 517 in so far as fully submerged
concrete is concerned. For other conditions of exposure, the danger of ingress of
chlorides influences the choice of cement, and therefore this topic is discussed in the
section dealing with chloride attack, in Chapter 11.
Disruption by alkali–silica reaction
In Chapter 3, the reactions between the alkalis and reactive silica and some carbonates
in aggregate were discussed. The consequences of the alkali–silica reaction and the
means of avoiding these consequences will now be considered.
The reaction can be disruptive and manifest itself as cracking. The crack width can
range from 0.1 mm to as much as 10 mm in extreme cases. The cracks are rarely more
than 25 mm, or at most 50 mm, deep.10.136 Hence, in most cases, the alkali–silica
reaction adversely affects the appearance and serviceability of a structure, rather than
its integrity; in particular, the compressive strength of concrete in the direction of the
applied stress is not greatly affected.10.115 Nevertheless, cracking can facilitate the
ingress of harmful agents.
The pattern of surface cracking induced by the alkali–silica reaction is irregular,
somewhat reminiscent of a huge spider’s web. However, the pattern is not necessarily
distinguishable from that caused by sulfate attack or by freezing and thawing, or even
by severe plastic shrinkage. In order to ascertain whether any observed cracking is
due to the alkali–silica reaction, a procedure recommended by a working party of the
British Cement Association10.112 can be followed. Within the concrete, many of the
cracks caused by the reaction can be seen to pass through individual aggregate
particles but also through the surrounding hydrated cement paste.
If the sole source of alkalis in concrete is Portland cement, then limiting the alkali
content in the cement would prevent the occurrence of deleterious reactions. The
minimum alkali content of cement at which expansive reaction can take place is 0.6
per cent of the soda equivalent. This is calculated from stoichiometry as the actual
Na2O content plus 0.658 times the K2O content of the clinker. This method of
calculation of the alkali content, which does not distinguish between sodium and
potassium, is convenient but simplistic. Chatterji10.119 found that potassium ions are
transported towards the silica faster than the sodium ions, and are therefore, on a mass
for mass basis, potentially more harmful.
The equivalent soda limit of 0.6 per cent by mass of cement lies at the origin
of low-alkali cement (see p. 48) and, indeed, defines it. It can be noted, nevertheless,
that, in exceptional cases, cements with an even lower alkali content have been known
to cause expansion.10.1
The background to the low-alkali cement, offered by Hobbs,10.128 may be of interest.
The alkali–silica reaction takes place only at high concentrations of OH–, that is at
high values of pH in the pore water. Now, the pH of the pore water depends on the
alkali content of the cement. Specifically, a high-alkali cement leads to a pH of
between 13.5 and 13.9 while a low-alkali cement results in a pH of 12.7 to
13.1.10.128 Given that an increase in pH of 1.0 represents a ten-fold increase in hydroxyl

562
ion concentration, the hydroxyl ion concentration with a low-alkali cement is about 10
times lower than when a high-alkali cement is used. This is the rationale of using low-
alkali cement with potentially reactive aggregate.
The assumption of prevention of a deleterious alkali–silica reaction by limiting the
alkali content in cement is valid only when two conditions are satisfied: there is no
other source of alkalis in the concrete; and the alkalis do not become concentrated in
some locations, at the expense of others. Such concentration may be caused by
moisture gradients or by alternating wetting and drying.10.118 This may be an
appropriate place to mention that the alkalis may also become concentrated by an
electric current passed through the concrete; this may occur when cathodic protection
is used to prevent corrosion of embedded steel.10.114
The additional sources of alkalis in concrete include sodium chloride present in
unwashed sand dredged from the sea or obtained from the desert. The use of such
sand in reinforced concrete should not be allowed because chlorides are conducive to
corrosion of steel (see Chapter 11). Other internal sources of alkalis are some
admixtures, especially superplasticizers, or even the mix water. The alkalis from these
sources, and also from fly ash and ground granulated blastfurnace slag, should be
included in the calculation of the amount of alkalis present, but only taking a
proportion of the actual amount of alkalis in these cementitious materials. There is no
agreement on how much this proportion should be, but BS 5328 : 4 : 1990 (replaced
by BS EN 206-1-2000) uses 17 per cent for fly ash and 50 per cent for ground
granulated blastfurnace slag.
Because of the varied provenance of alkalis, it is logical to limit the total content of
alkalis in concrete. British Standard BS 5328-1 : 1991 (withdrawn) specifies a
maximum of 3.0 kg of alkalis (expressed as soda equivalent) which can be present in
1 m3 of concrete containing alkali-reactive aggregate. This amount of reactive alkalis
is determined by a British method which is different from that prescribed by British
Standard BS EN 196-21 : 1992 (withdrawn); the latter method gives a value for the
alkali content about 0.025 percentage point higher than the British method. Thus,
when compliance with BS 5328-1 : 1997 (withdrawn) is required, it is important to
exercise care in the choice of the test method for the determination of the alkali
content of cement.
It is useful to reiterate the fact that there are three necessary conditions for alkali-
silica reaction to proceed: alkalis, reactive silica and adequate moisture. The latter is
taken usually as not less than 85 per cent relative humidity (which is common
outdoors in the UK at night or in winter and which may be present in the interior of
concrete members due to residual mixing water). Alkalis are always present in
Portland cement, but there exists low-alkali Portland cement (see p. 48). In addition,
BRE Digest 330.3 recognizes also moderate alkali and high alkali cements. Alkalis
are sometimes also present in admixtures. The British Standards of relevance are BS
EN 206-1 : 2000 and BS 8500-2 : 2002.
From the above, it follows that aggregates with a low alkali reactivity should be
used, and in the UK they are usually available. However, NaCl is a source of alkalis,
and residual NaCl may be present in poorly washed sea-dredged sand and also in de-
icing salts.
In practice, it is not easy to eliminate totally the alkali-silica reaction, but we
should aim to minimize that reaction. That topic is discussed in ref. 10.141. However,
assistance in minimizing the alkali-silica reaction is obtained by incorporating ggbs or

563
fly ash with a minimum of 25 percent of the mass of cement. There is a paradox here
because both ggbs and fly ash contain alkalis but these are in glass form.
Preventive measures
The discussion of the alkali–silica reaction, presented in Chapter 3, makes it clear that
the progress and consequences of the reaction are influenced by the proportions of
various ions in the pore water and by the availability of the alkalis and of silica. In
particular, the expansion caused by the alkali–silica reaction is greater the greater the
content of reactive silica, but only up to a certain content of silica; at higher contents,
the expansion is smaller. This is illustrated in Fig. 10.10.10.6 There is thus a pessimum
content of silica. This pessimum content is higher at lower water/cement ratios and at
higher cement contents.10.128 The ratio of reactive silica to the alkalis corresponding to
the maximum expansion usually lies in the range of 3.5 to 5.5.10.128

Fig. 10.10. Relation between expansion after 224 days and reactive silica content
in the aggregate10.6
It follows from the above that varying the silica content in concrete can move the
silica/alkali ratio away from the pessimum. Specifically, it has been found that
expansion due to the alkali–silica reaction can be reduced or eliminated by the
addition to the mix of reactive silica in a finely powdered form. This apparent paradox
can be explained by reference to Fig. 10.10, showing the relation between the
expansion of a mortar bar and the content of reactive silica of size between 850 and
300 μm (No. 20 and No. 50 ASTM) sieves, i.e. not in a powdered form.10.6 In the range
of low silica contents, the greater quantity of silica for a given amount of alkalis
increases expansion but, with higher values of silica content, the situation is reversed:
the greater the surface area of the reactive aggregate the lower the quantity of alkalis
available per unit of this area, and the less alkali–silica gel can be formed.10.6 On the
other hand, due to the extremely low mobility of calcium hydroxide, only that

564
adjacent to the surface of the aggregate is available for reaction, so that the quantity of
calcium hydroxide per unit area of aggregate is independent of the magnitude of the
total surface area of the aggregate. Thus, increasing the surface area increases the
calcium hydroxide/alkali ratio of the solution at the boundary of the aggregate. Under
such circumstances, an innocuous (non-expanding) calcium alkali silicate product is
formed.10.8
By a similar argument, finely divided siliceous material added to the coarse
reactive particles already present would reduce expansion, although the reaction with
the alkalis still takes place. These pozzolanic additions, such as crushed pyrex glass or
fly ash, have indeed been found effective in reducing the penetration of the coarser
aggregate particles. The fly ash should contain no more than 2 or 3 per cent by mass
of alkalis.10.136 However, Class F fly ash, used in a quantity representing 58 per cent (by
mass) of the total cementitious material, was found to be highly effective in
preventing expansion, even when the total alkali content was 5 kg per cubic metre of
concrete.10.117 It is important that the fly ash be fine; if necessary, grinding can be used
to improve its effectiveness in reducing the expansion.
Pozzolanas in the mix are beneficial also because they reduce the permeability of
concrete (see Chapter 13) and therefore reduce the mobility of aggressive agents, both
those present within the concrete and those which may ingress. Furthermore, C-S-H
formed by pozzolanic activity incorporates a certain amount of alkalis and thus lowers
the value of pH;10.136 the influence of pH upon the alkali–silica reaction was discussed
earlier in this section.
Silica fume is particularly effective because the silica reacts preferentially with the
alkalis. Although the product of reaction is the same as that between the alkalis and
the reactive silica in the aggregate, the reaction takes place at the very large surface of
the fine particles of silica fume (see p. 87). In consequence, the reaction does not
result in expansion.10.116
Ground granulated blastfurnace slag is also effective in mitigating or preventing
the deleterious effects of the alkali–silica reaction. It should be noted that the presence
of ground granulated blastfurnace slag results in a reduced permeability of concrete
(see Chapter 13). There is evidence that, when Portland blastfurnace slag cement is
used, a maximum alkali content of 0.9 per cent is harmless when the slag content of
the cement is not less than 50 per cent.10.99 An even higher alkali content, 1.1 per cent,
is considered to be tolerable by BS 5328-1 : 1991, replaced by BS EN 206-1 : 2000.
There is anecdotal evidence of the beneficial effects of ground granulated blastfurnace
slag in so far as the deleterious alkali–silica expansion is concerned. In the
Netherlands, deleterious expansion in a number of structures was observed but it was
absent where Portland blastfurnace cement had been used.10.122
To be effective, the various cementitious materials must be present in adequate
proportions of the total cementitious material. Expressed by mass, these proportions
are as follows: Class F fly ash–at least 30 or 40 per cent; silica fume–at least 20 per
cent; ground granulated blastfurnace slag–50 to 65 per cent.10.120,10.136 Inadequate
amounts can actually aggravate the situation and increase expansion if a particularly
bad silica–alkali ratio is reached (cf. Fig. 10.10). The performance of any pozzolana
or ground granulated blastfurnace slag in preventing excessive expansion due to
alkali–silica reaction should be tested according to ASTM C 441-05. Advice
contained in an appendix to Canadian Standard A23.1-94 is very useful.10.111

565
 Inclusion in the concrete of silica fume or fly ash will not be effective in
preventing expansion if alkalis can continue to ingress into the concrete.10.113,10.119 More
generally, in considering the alkali content in concrete, it should be noted that water-
borne alkalis can ingress from outside in some structures, for example, from other,
adjacent building materials or from sodium chloride used as a de-icing agent.
Some tests indicate that lithium salts may inhibit expansive reactions but the
relevant mechanism has not been established.10.121
It should be noted that, although silica gel resulting from the alkali-silica reaction
can be formed inside air bubbles, it does not follow that air entrainment represents a
means of avoiding the deleterious effects of the reaction.
Abrasion of concrete
Under many circumstances, concrete surfaces are subjected to wear. This may be due
to attrition by sliding, scraping or percussion.10.14 In the case of hydraulic structures, the
action of the abrasive materials carried by water leads to erosion. Another cause of
damage to concrete in flowing water is cavitation.
Tests for abrasion resistance
Resistance of concrete to abrasion is difficult to assess because the damaging action
varies depending on the exact cause of wear, and no one test procedure is satisfactory
in evaluating all the conditions: rubbing test, including rolling balls, dressing wheel,
or sandblasting may each be appropriate in different cases.
ASTM C 418-05 prescribes the procedure for determining wear by sandblasting;
the loss of volume of concrete serves as a basis for judgement, but not as a criterion of
wear resistance under different conditions. ASTM C 779-05 prescribes three test
procedures for laboratory or field use. In the revolving disc test, there is applied a
revolving motion of three flat surfaces driven along a circular path at 0.2 Hz and
individually turning on their axes at 4.6 Hz. Silicon carbide is fed as an abrasive
material. In the steel ball abrasion test, a load is applied to a rotating head which is
separated from the specimen by steel balls. The test is performed in circulating water
in order to remove the eroded material. The dressing wheel test uses a drill press
modified to apply a load to three sets of seven rotating dressing wheels which are in
contact with the specimen. The driving head is rotated for 30 minutes at 0.92 Hz. In
all cases, the depth of wear of the specimen is used as a measure of abrasion.
When it is desired to perform abrasion tests on cores (which are too small for the
tests of ASTM C 418-05 and C 779-05) ASTM C 944-99 (2005) can be used. Here,
two dressing wheels in a drill press under a fixed load are applied to the surface of the
core, and the loss of mass is determined; the depth of wear can also be measured.
The various tests try to simulate the modes of abrasion found in practice, but this is
not easy and, indeed, the main difficulty in abrasion testing is to make sure that the
results of a test represent the comparative resistance of concrete to a given type of
wear. The tests prescribed by ASTM 779-05 are useful in estimating the resistance of
concrete to heavy foot and wheeled traffic and to tyre chains and tracked vehicles.
Broadly speaking, the heavier the abrasion in service the more helpful the test in
increasing order of usefulness: revolving disc, dressing wheel, and steel ball.10.32
Figure 10.11 shows the results of the three tests of ASTM 779-05 on different
concretes. Because of the arbitrary conditions of test, the values obtained are not
comparable quantitatively, but in all cases the resistance to abrasion was found to be

566
proportional to the compressive strength of concrete.10.20 The steel ball test appears
more consistent and more sensitive than the other tests.

Fig. 10.11. Influence of the water/cement ratio of the mix on the abrasion loss of
concrete for different tests10.20
The resistance of concrete to abrasion by water-borne solids can be determined
using ASTM C 1138-05. In this test, the behaviour of swirling water containing
suspended particles is simulated by high-speed movement of steel grinding balls of
various sizes in a water tank over a period of 72 hours. The depth of wear of the
concrete surface gives a comparative measurement.
In a totally different category of assessing the abrasion resistance of concrete is the
rebound hammer test (see p. 626): the value obtained is sensitive to some of the
factors which influence the abrasion resistance of concrete.10.37
Factors influencing abrasion resistance
Abrasion seems to involve high-intensity stress applied locally so that the strength and
hardness of the surface zone of concrete strongly influence the resistance to abrasion.

567
In consequence, the compressive strength of concrete is the principal factor
controlling the resistance to abrasion. The minimum strength required depends on the
severity of abrasion expected. Very high strength concretes exhibit a high resistance
to abrasion: for example, increasing the compressive strength from 50 MPa (7000 psi)
to 100 MPa (14 000 psi) increases the abrasion resistance by 50 per cent, and 150
MPa (21 000 psi) concrete is as resistant as high-quality granite.10.40
The properties of the concrete in the surface zone are strongly affected by the
finishing operations, which may reduce the water/cement ratio and improve
compaction. Vacuum dewatering is beneficial (see p. 234). The presence of laitance
must be avoided. Particularly good curing is of importance; it is desirable to have a
period of curing twice as long as normal in order to achieve good resistance to
abrasion.
Rich mixes are undesirable, a cement content of 350 kg/m3 (600 lb/yd3) being
probably a maximum because coarse aggregate should be present just below the
surface of the concrete.
As far as the aggregate is concerned, inclusion of some crushed sand is
desirable,10.40 and so is the use of strong10.38 and hard aggregate; however, the abrasion
resistance of aggregate, as determined by the Los Angeles test (see p. 124) does not
seem to be a good indicator of the abrasion resistance of concrete made with a given
aggregate.10.39 Lightweight aggregate of high quality has good abrasion resistance
because it is inherently a ceramic material but, due to its porous structure, it is not
resistant to impact which may be associated with abrasion.10.87
Shrinkage-compensating concrete has a significantly increased abrasion
resistance10.94 probably because of the absence of fine cracks which would encourage
the progress of abrasion.
Consideration of the use of hardeners incorporated in the surface zone of concrete
is outside the scope of the present book.
Erosion resistance
Erosion of concrete is an important type of wear which may occur in concrete in
contact with flowing water. It is convenient to distinguish between erosion due to
solid particles carried by the water and damage due to pitting resulting from cavities
forming and collapsing in water flowing at high velocities. The latter is considered in
the next section.
The rate of erosion depends on the quantity, shape, size, and hardness of the
particles being transported, on the velocity of their movement, on the presence of
eddies, and also on the quality of concrete.10.41 As in the case of abrasion in general,
this quality appears to be best measured by the compressive strength of concrete but
the mix composition is also relevant. In particular, concrete with large aggregate
erodes less than mortar of equal strength, and hard aggregate improves the erosion
resistance. However, under some conditions of wear, smaller-size aggregate leads to a
more uniform erosion of the surface. In general, at a constant slump, the erosion
resistance increases with a decrease in the cement content;10.15 this has the advantage of
reducing laitance. At a constant cement content, the resistance improves with a
decrease in slump:10.15 this is probably in agreement with the general influence of
compressive strength.

568
 In all cases, of course, it is only the quality of the concrete in the surface zone that
is relevant, but even the best concrete will rarely withstand severe erosion over
prolonged periods. Vacuum dewatering and use of permeable formwork are beneficial.
Proneness to erosion by solids in flowing water can be measured by means of
a shot-blast test. Here, 2000 pieces of broken steel shot (of 850 μm (No. 20 ASTM)
sieve size) are ejected under air pressure of 0.62 MPa (90 psi) from a 6.3 mm ( in.)
nozzle against a concrete specimen 102 mm (4 in.) away.
Cavitation resistance
While good quality concrete can withstand steady, tangential, high-velocity flow of
water, severe damage rapidly occurs in the presence of cavitation. By this is meant the
formation of vapour bubbles when the local absolute pressure drops to the value of the
ambient vapour pressure of water at the ambient temperature. The bubbles or cavities
can be large, single voids, which later break up, or clouds of small bubbles.10.16 They
flow downstream and, on entering an area of higher pressure, collapse with great
impact. Because the collapse of the cavities means entry of high-velocity water into
the previously vapour-occupied space, extremely high pressure on a small area is
generated during very short time intervals, and it is the repeated collapse over a given
part of the concrete surface that causes pitting. Greatest damage is caused by clouds
of minute cavities found in eddies. They usually coalesce momentarily into a large
amorphous cavity which collapses extremely rapidly.10.17 Many of the cavities pulsate
at a high frequency, and this seems to aggravate damage over an extended area.10.18
Cavitation damage occurs in open channels generally only at velocities in excess
of 12 m/s (40 ft/s),10.41 but in closed conduits even at much lower speeds when there is
a possibility of pressure dropping well below atmospheric. Such a drop may be caused
by syphonic action, or by inertia on the inside of a bend or over boundary
irregularities; often, there is a combination of these. Divergence of flow from the
concrete surface of an open channel is a frequent cause of cavitation. Although the
advent of cavitation depends primarily on pressure changes (and consequently also on
velocity changes), it is especially likely to occur in the presence of small quantities of
undissolved air in the water. These bubbles of air behave as nuclei at which the
change of phase from liquid to vapour can more readily occur. Dust particles have a
similar effect, possibly because they ‘house’ the undissolved air. On the other hand,
small bubbles of free air in large quantities (up to 8 per cent by volume near the
surface of concrete), while promoting cavitation, cushion the collapse of the cavities
and hence reduce the cavitation damage.10.19 Deliberate aeration of water may therefore
be advantageous.10.41
The surface of concrete affected by cavitation is irregular, jagged and pitted, in
contrast to the smoothly worn surface of concrete eroded by water-borne solids. The
cavitation damage does not progress steadily: usually, after an initial period of small
damage, rapid deterioration occurs, followed by damage at a slower rate.10.19
Best resistance to cavitation damage is obtained by the use of high strength
concrete, possibly formed by an absorptive lining (which reduces the local
water/cement ratio). The maximum size of aggregate near the surface should not
exceed 20 mm ( in.)10.19 because cavitation tends to remove large particles. Hardness
of aggregate is not important (unlike the case of erosion resistance) but good bond
between aggregate and mortar is vital.

569
 Use of polymers, steel fibres or resilient coatings may improve the cavitation
resistance, but these topics are outside the scope of the present book. However, while
the use of suitable concrete may reduce cavitation damage, not even the best concrete
can withstand cavitation forces for an indefinite time. The solution of the cavitation
damage problem lies, therefore, primarily in reducing cavitation. This can be achieved
by the provision of smooth and well-aligned surfaces free from irregularities such as
depressions, projections, joints and misalignments, and by the absence of abrupt
changes in slope or curvature that tend to pull the flow away from the surface. If
possible, local increase in velocity of water should be avoided as damage is
proportional to the sixth or seventh power of velocity.10.19
Types of cracking
Because cracking may impair the durability of concrete by allowing ingress of
aggressive agents, it is relevant to review briefly the types and causes of cracking. In
addition, cracking may adversely affect the watertightness or sound transmission of
structures or mar their appearance. With respect to appearance, the acceptable crack
width depends on the distance from which it is viewed and on the function of the
structure, e.g. a public hall, at one extreme, and a warehouse, at the other. It may be
useful to add that ingress of dirt makes the cracks more perceptible; so does the use of
white cement in concrete.
As far as water-tightness is concerned, very narrow, non-moving cracks, 0.12 to
0.20 mm (0.005 to 0.008 in.) wide, may leak initially.10.33,10.34 However, dissolved
calcium hydroxide carried by slowly percolating water may react with atmospheric
carbon dioxide to deposit calcium carbonate, which would seal the crack10.33 (see
p. 330).
Cracking occurring in fresh concrete, that is, plastic shrinkage cracking and plastic
settlement cracking, was discussed in Chapter 9. Another type of early cracking is
known as crazing, which can occur on slabs or walls when the surface zone of the
concrete has a higher water content than deeper in the interior. The pattern of crazing
looks like an irregular network with a spacing of up to about 100 mm (4 in.). The
cracks are very shallow and develop early, but may not be noticed until etched by dirt;
apart from appearance, they are of little importance.
In addition, a somewhat different kind of surface damage, known as blisters, can
occur if some bleed water or large air bubbles are trapped just below the surface of the
concrete by a thin layer of laitance induced by finishing. Blisters are 10 to 100 mm
( to 4 in.) in diameter and 2 to 10 mm (or to in.) thick. In service, the laitance
layer becomes detached, leaving behind a shallow depression.
In hardened concrete, cracking may be caused by drying shrinkage or by restrained
early-age thermal movement; these were discussed in Chapters 9 and 8, respectively.
The various types of non-structural cracks are listed in Table 10.8 and shown
schematically in Fig. 10.12.10.33 It is useful to note that, whereas one particular cause
may initiate a crack, its development can be due to another cause.10.33 Thus a diagnosis
of causes of cracking is not always straightforward.

Table 10.8. Classification of Intrinsic Cracks (based on ref. 10.33)

570
571
572
 Fig. 10.12. Schematic representation of the various types of cracking which can
occur in concrete (see Table 10.8) (based on ref. 10.33)
Cracking can also be caused by overloading in relation to the actual strength of the
concrete member, but this is the consequence of inadequate design, or of construction
not conforming to the specification. It is important to remember that, in reinforced
concrete in service, tension is induced in the reinforcing steel and in the surrounding
concrete. Surface cracking is therefore inevitable but, with proper structural design
and detailing, the cracks are very narrow and barely perceptible. Stress-induced
cracks have a maximum width at the surface of the concrete and taper towards the
steel, but the difference in width may decrease with time.10.34 The crack width at the
surface is greater the larger the cover to reinforcement.
We should note that, from energy considerations, it is easier to extend an existing
crack than to form a new one. This explains why, under an applied load, each
subsequent crack occurs under a higher load than the preceding one. The total number
of cracks developed is determined by the size of the concrete member, and the
distance between cracks depends on the maximum size of aggregate present.10.106
Because, under given physical conditions, the total crack width per unit length of
concrete is fixed and we want the cracks to be as fine as possible, it is desirable to
have more cracks. For this reason, the restraint to cracking should be uniform along
the length of the member. Provision of reinforcement controls shrinkage cracking by
reducing the width of individual cracks, but not the total width of all the cracks taken
together. This topic is outside the scope of the present book.

573
 The importance of cracking, and the minimum width at which a crack is
considered significant, depend on the function of the structural members and on the
conditions of exposure of the concrete. Reis et al.10.105 suggested the following
permissible crack widths, which still offer good guidance:
Interior members 0.35 mm (0.014
in.)
Exterior members under normal exposure conditions 0.25 mm (0.010
in.)
Exterior members exposed to particularly aggressive environment 0.15 mm
(0.006 in.).
It may be relevant to mention that, although there is a variation between observers,
the minimum crack width that can be seen with a naked eye is about 0.13 mm (0.005
in.). Simple magnifying devices make it possible to determine the crack width.
Various specialized techniques, such as electro-conductive paint and light-dependent
resistors, make it possible to determine the development of cracking. However, very
fine cracks are very common but not harmful, so that intensive searching for cracks
serves no purpose.
References
10.1. W. C. HANNA, Additional information on inhibiting alkali–aggregate
expansion, J. Amer. Concr. Inst., 48, p. 513 (Feb. 1952).
10.2. T. C. POWERS, H. M. MANN and L. E. COPELAND, The flow of water in hardened
portland cement paste, Highw. Res. Bd Sp. Rep. No. 40, pp. 308–23
(Washington DC, July 1959).
10.3. T. C. POWERS, Structure and physical properties of hardened portland cement
paste, J. Amer. Ceramic Soc., 41, pp. 1–6 (Jan. 1958).
10.4. T. C. POWERS, L. E. COPELAND and H. M. MANN, Capillary continuity or
discontinuity in cement pastes, J. Portl. Cem. Assoc. Research and
Development Labortories, 1, No. 2, pp. 38–48 (May 1959).
10.5. T. C. POWERS, L. E. COPELAND, J. C. HAYES and H. M. MANN, Permeability of
Portland cement paste, J. Amer. Concr. Inst., 51, pp. 285–98 (Nov. 1954).
10.6. H. E. VIVIAN, Studies in cement–aggregate reaction: X. The effect on mortar
expansion of amount of reactive component, Commonwealth Scientific and
Industrial Research Organization Bull. No. 256, pp. 13–20 (Melbourne,
1950).
10.7. F. M. LEA, The Chemistry of Cement and Concrete (London, Arnold, 1970).
10.8. G. J. VERBECK and C. GRAMLICH, Osmotic studies and hypothesis concerning
alkali–aggregate reaction, Proc. ASTM, 55, pp. 1110–28 (1955).
10.9. J. H. P. VAN AARDT, The resistance of concrete and mortar to chemical attack –
progress report on concrete corrosion studies, National Building Research
Institute, Bull. No. 13, pp. 44–60 (South African Council for Scientific and
Industrial Research, March 1955).
10.10. J. H. P. VAN AARDT, Chemical and physical aspects of weathering and
corrosion of cement products with special reference to the influence of warm
climate, RILEM Symposium on Concrete and Reinforced Concrete in Hot
Countries (Haifa, 1960).

574
 10.11. L. H. TUTHILL, Resistance to chemical attack, ASTM Sp. Tech. Publ. No. 169,
pp. 188–200 (1956).
10.12. R. L. HENRY and G. K. KURTZ, Water vapor transmission of concrete and of
aggregates. U.S. Naval Civil Engineering Laboratory, Port Hueneme,
California, 71 pp. (June 1963).
10.13. A. M. NEVILLE, Behaviour of concrete in saturated and weak solutions of
magnesium sulphate and calcium chloride, J. Mat., ASTM, 4, No. 4, pp. 781–
816 (Dec. 1969).
10.14. M. E. PRIOR, Abrasion resistance, ASTM Sp. Tech. Publ. No. 169A, pp. 246–
60 (1966).
10.15. U.S. ARMY CORPS OF ENGINEERS, Concrete abrasion study, Bonneville Spillway
Dam, Report 15-1 (Bonneville, Or., Oct. 1943).
10.16. J. M. HOBBS, Current ideas on cavitation erosion, Pumping, 5, No. 51, pp.
142–9 (March 1963).
10.17. M. J. KENN, Cavitation eddies and their incipient damage to concrete, Civil
Engineering, 61, No. 724, pp. 1404–5 (London, Nov. 1966).
10.18. S. P. KOZIREV, Cavitation and cavitation-abrasive wear caused by the flow of
liquid carrying abrasive particles over rough surfaces, Translation by The
British Hydro-mechanics Research Association (Feb. 1965).
10.19. M. J. KENN, Factors influencing the erosion of concrete by cavitation, CIRIA,
15 pp. (London, July 1968).
10.20. F. L. SMITH, Effect of aggregate quality on resistance of concrete to
abrasion, ASTM Sp. Tech. Publ. No. 205, pp. 91–105 (1958).
10.21. J. BONZEL, Der Einfluss des Zements, des W/Z Wertes, des Alters und der
Lagerung auf die Wasserundurchlässigkeit des Betons, Beton, No. 9, pp.
379–83; No. 10, pp. 417–21 (1966).
10.22. J. W. FIGG, Methods of measuring the air and water permeability of
concrete, Mag. Concr. Res., 25, No. 85, pp. 213–19 (Dec. 1973).
10.23. P. J. SEREDA and V. S. RAMACHANDRAN, Predictability gaps between science and
technology of cements – 2, Physical and mechanical behavior of hydrated
cements. J. Amer. Ceramic Soc., 58, Nos 5–6, pp. 249–53 (1975).
10.24. G. J. OSBORNE and M. A. SMITH, Sulphate resistance and long-term strength
properties of regulated-set cements, Mag. Concr. Res., 29, No. 101, pp. 213–
24 (1977).
10.25. B. MATHER, How soon is soon enough?, J. Amer. Concr. Inst., 73, No. 3, pp.
147–50 (1976).
10.26. L. ROMBÈN, Aspects of testing methods for acid attack on concrete. CBI
Research, 1: 78, 61 pp. (Swedish Cement and Concrete Research Inst.,
1978).
10.27. H. T. THORNTON, Acid attack of concrete caused by sulfur bacteria action, J.
Amer. Concr. Inst., 75, No. 11, pp. 577–84 (1978).
10.28. H. U. CHRISTEN, Conditions météorologiques et efflorescences de
chaux, Bulletin du Ciment, 44, No. 6, 8 pp. (Wildegg, Switzerland, June
1976).

575
 10.29. BULLETIN DU CIMENT, Traitement des surfaces de béton à l’acide, 45, No. 21, 6
pp. (Wildegg, Switzerland, Sept. 1977).
10.30. BULLETIN DU CIMENT, Coloration sombre du béton, 45, No. 23, 6 pp. (Wildegg,
Switzerland, Nov. 1977).
10.31. L. H. TUTHILL, Resistance to chemical attack, ASTM Sp. Tech. Publ. No. 169B,
pp. 369–87 (1978).
10.32. R. O. LANE, Abrasion resistance, ASTM Sp. Tech. Publ. No. 169B, pp. 332–
50 (1978).
10.33. CONCRETE SOCIETY REPORT, Non-structural Cracks in Concrete, Technical
Report No. 22, 4th Edn, 62 pp. (London, Concrete Society, 2010).
10.34. ACI 207.2R-90, Effect of restraint, volume change, and reinforcement on
cracking of mass concrete, ACI Manual of Concrete Practice, Part 1:
Materials and General Properties of Concrete, 18 pp. (Detroit, Michigan,
1994).
10.35. V. G. PAPADAKIS, M. N. FARDIS and C. G. VAYENAS, Effect of composition,
environmental factors and cement-lime mortar coating on concrete
carbonation, Materials and Structures, 25, No. 149, pp. 293–304 (1992).
10.36. D. W. S. Ho and R. K. LEWIS, The specification of concrete for reinforcement
protection – performance criteria and compliance by strength, Cement and
Concrete Research, 18, No. 4, pp. 584–94 (1988).
10.37. M. SADEGZADEH and R. KETTLE, Indirect and non-destructive methods for
assessing abrasion resistance of concrete, Mag. Concr. Res., 38, No. 137, pp.
183–90 (1986).
10.38. P. LAPLANTE, P.-C. AÏTCIN and D. VÉZINA, Abrasion resistance of
concrete, Journal of Materials in Civil Engineering, 3, No. 1, pp. 19–28
(1991).
10.39. T. C. LIU, Abrasion resistance of concrete, ACI Journal, 78, No. 5, pp. 341–
50 (1981).
10.40. O. E. GJØRV, T. BAERLAND and H. H. RONNING, Increasing service life of
roadways and bridges, Concrete International, 12, No. 1, pp. 45–8 (1990).
10.41. ACI 210R-93, Erosion of concrete in hydraulic structures, ACI Manual of
Concrete Practice, Part 1: Materials and General Properties of Concrete,
24 pp. (Detroit, Michigan, 1994).
10.42. ACI 201.2R-1992, Guide to durable concrete, ACI Manual of Concrete
Practice, Part 1: Materials and General Properties of Concrete, 41 pp.
(Detroit, Michigan, 1994).
10.43. U.S. BUREAU OF RECLAMATION, 4913-92, Procedure for water permeability of
concrete, Concrete Manual, Part 2, 9th Edn, pp. 714–25 (Denver, Colorado,
1992).
10.44. J. F. YOUNG, A review of the pore structure of cement paste and concrete and
its influence on permeability, in Permeability of Concrete, ACI SP-108, pp.
1–18 (Detroit, Michigan, 1988).
10.45. A. BISAILLON and V. M. MALHOTRA, Permeability of concrete: using a uniaxial
water-flow method, in Permeability of Concrete, ACI SP-108, pp. 173–93
(Detroit, Michigan, 1988).

576
 10.46. R. A. COOK and K. C. HOVER, Mercury porosimetry of cement-based materials
and associated correction factors, ACI Materials Journal, 90, No. 2, pp.
152–61 (1993).
10.47. J. VUORINEN, Applications of diffusion theory to permeability tests on
concrete Part I: Depth of water penetration into concrete and coefficient of
permeability, Mag. Concr. Res., 37, No. 132, pp. 145–52 (1985).
10.48. O. VALENTA, Kinetics of water penetration into concrete as an important
factor of its deterioration and of reinforcement corrosion, RILEM
International Symposium on the Durability of Concrete, Prague, Part I, pp.
177–93 (1969).
10.49. L. A. LARBI, Microstructure of the interfacial zone around aggregate particles
in concrete, Heron, 38, No. 1, 69 pp. (1993).
10.50. A. HANAOR and P. J. E. SULLIVAN, Factors affecting concrete permeability to
cryogenic fluids, Mag. Concr. Res., 35, No. 124, pp. 142–50 (1983).
10.51. D. WHITING, Permeability of selected concretes, in Permeability of Concrete,
ACI SP-108, pp. 195–221 (Detroit, Michigan, 1988).
10.52. C. D. LAWRENCE, Transport of oxygen through concrete, in The Chemistry and
Chemically-Related Properties of Cement, Ed. F. P. Glasser, British Ceramic
Proceedings, No. 35, pp. 277–93 (1984).
10.53. J. J. KOLLEK, The determination of the permeability of concrete to oxygen by
the Cembureau method – a recommendation, Materials and Structures, 22,
No. 129, pp. 225–30 (1989).
10.54. RILEM RECOMMENDATIONS CPC-18, Measurement of hardened concrete
carbonation depth, Materials and Structures, 21, No. 126, pp. 453–5 (1988).
10.55. L. J. PARROTT, A Review of Carbonation in Reinforced Concrete, Cement and
Concrete Assn, 42 pp. (Slough, U.K., July 1987).
10.56. V. G. PAPADAKIS, C. G. VAYENAS and M. N. FARDIS, Fundamental modeling and
experimental investigation of concrete carbonation, ACI Materials
Journal, 88, No. 4, pp. 363–73 (1991).
10.57. M. SOHUI, Case study on durability, Darmstadt Concrete, 3, pp. 199–207
(1988).
10.58. R. J. CURRIE, Carbonation depths in structural-quality concrete. Building
Research Establishment Report, 19 pp. (Watford, U.K., 1986).
10.59. L. TANG and L.-O. NILSSON, Effect of drying at an early age on moisture
distributions in concrete specimens used for air permeability test, in Nordic
Concrete Research, Publication 13/2/93, pp. 88–97 (Oslo, Dec. 1993).
10.60. G. K. MOIR and S. KELHAM, Durability 1, Performance of Limestone-filled
Cements, Proc. Seminar of BRE/BCA Working Party, pp. 7.1–7.8 (Watford,
U.K. 1989).
10.61. L. J. PARROTT and D. C. KILLOCH, Carbonation in 36 year old, in-situ
concrete, Cement and Concrete Research, 19, No. 4, pp. 649–56 (1989).
10.62. P. NISCHER, Einfluss der Betongüte auf die Karbonatisierung, Zement und
Beton, 29, No. 1, pp. 11–15 (1984).
10.63. M. D. A. THOMAS and J. D. MATTHEWS, Carbonation of fly ash concrete, Mag.
Concr. Res., 44, No. 160, pp. 217–28 (1992).

577
 10.64. G. J. OSBORNE, Carbonation of blastfurnace slag cement concretes, Durability
of Building Materials, 4, pp. 81–96 (Amsterdam, Elsevier Science, 1986).
10.65. K. HORIGUCHI et al., The rate of carbonation in concrete made with blended
cement, in Durability of Concrete, ACI SP-145, pp. 917–31 (Detroit,
Michigan, 1994).
10.66. D. W. HOBBS, Carbonation of concrete containing pfa, Mag. Concr. Res., 40,
No. 143, pp. 69–78 (1988).
10.67. Th. A. BIER, Influence of type of cement and curing on carbonation progress
and pore structure of hydrated cement paste, Materials Research Society
Symposium, 85, pp. 123–34 (1987).
10.68. RILEM RECOMMENDATIONS TC 71-PSL, Systematic methodology for service
life. Prediction of building materials and components, Materials and
Structures, 22, No. 131, pp. 385–92 (1988).
10.69. M. C. MCVAY, L. D. SMITHSON and C. MANZIONE, Chemical damage to airfield
concrete aprons from heat and oils, ACI Materials Journal, 90, No. 3, pp.
253–8 (1993).
10.70. H. L. KONG and J. G. ORBISON, Concrete deterioration due to acid
precipitation, ACI Materials Journal, 84, No. 3, pp. 110–16 (1987).
10.71. I. BICZOK, Concrete Corrosion and Concrete Protection, 8th Edn, 545 pp.
(Budapest, Akadėmiai Kiad , 1972).
10.72. P. D. CARTER, Preventive maintenance of concrete bridge decks, Concrete
International, 11, No. 11, pp. 33–6 (1989).
10.73. M. R. SILVA and F.-X. DELOYE, Dégradation biologique des bétons, Bulletin
Liaison Laboratoires Ponts et Chausseés, 176, pp. 87–91 (Nov.–Dec. 1991).
10.74. R. DRON and F. BRIVOT, Le gonflement ettringitique, Bulletin Liaison
Laboratoires Ponts et Chausseés, 161, pp. 25–32 (May–June 1989).
10.75. I. ODLER and M. GLASSER, Mechanism of sulfate expansion in hydrated
portland cement, J. Amer. Ceramic Soc., 71, No. 11, pp. 1015–20 (1988).
10.76. K. MATHER, Factors affecting sulfate resistance of mortars, Proceedings 7th
International Congress on Chemistry of Cement, Paris, Vol. IV, pp. 580–5
(1981).
10.77. P. J. TIKALSKY and R. L. CARRASQUILLO, Influence of fly ash on the sulfate
resistance of concrete. ACI Materials Journal, 89, No. 1, pp. 69–75 (1992).
10.78. N. I. FATTUHI and B. P. HUGHES, The performance of cement paste and
concrete subjected to sulphuric acid attack, Cement and Concrete
Research, 18, No. 4, pp. 545–53 (1988).
10.79. K. R. LAUER, Classification of concrete damage caused by chemical attack,
RILEM Recommendation 104-DDC: Damage Classification of Concrete
Structures, Materials and Structures, 23, No. 135, pp. 223–9 (1990).
10.80. G. J. OSBORNE, The sulphate resistance of Portland and blastfurnace slag
cement concretes, in Durability of Concrete, Vol. II, Proceedings 2nd
International Conference, Montreal, ACI SP-126, pp. 1047–61 (1991).
10.81. B. MATHER, A discussion of the paper “Theories of expansion in
sulfoaluminate-type expansive cements: schools of thought,” by M. D.
Cohen, Cement and Concrete Research, 14, pp. 603–9 (1984).

578
 10.82. V. A. ROSSETTI, G. CHIOCCHIO and A. E. PAOLINI, Expansive properties of the
mixture C4A H12-2C , III. Effects of temperature and restraint. Cement and
Concrete Research, 13, No. 1, pp. 23–33 (1983).
10.83. P. K. MEHTA, Sulfate attack on concrete – a critical review, Materials Science
of Concrete III, Ed. J. Skalny, American Ceramic Society, pp. 105–30
(1993).
10.84. M. L. CONJEAUD, Mechanism of sea water attack on cement mortar,
in Performance of Concrete in Marine Environment, ACI SP-65, pp. 39–61
(Detroit, Michigan, 1980).
10.85. K. MATHER, Concrete weathering at Treat Island, Maine, in Performance of
Concrete in Marine Environment, ACI SP-65, pp. 101–11 (Detroit,
Michigan, 1980).
10.86. O. E. GJØRV and O. VENNESLAND, Sea salts and alkalinity of concrete, ACI
Journal, 73, No. 9, pp. 512–16 (1976).
10.87. R. E. PHILLEO, Report of materials working group, Proceedings of
International Workshop on the Performance of Offshore Concrete Structures
in the Arctic Environment, National Bureau of Standards, pp. 19–25
(Washington DC, 1983).
10.88. B. MATHER, Effects of seawater on concrete, Highway Research Record, No.
113, Highway Research Board, pp. 33–42 (1966).
10.89. A. M. PAILLIÈRE et al., Influence of curing time on behaviour in seawater of
high-strength mortar with silica fume, in Durability of Concrete, ACI SP-
126, pp. 559–75 (Detroit, Michigan, 1991).
10.90. A. M. PAILLIÈRE, M. RAVERDY and J. J. SERRANO, Long term study of the
influence of the mineralogical composition of cements on resistance to
seawater: tests in artificial seawater and in the Channel, in Durability of
Concrete, ACI SP-145, pp. 423–43 (Detroit, Michigan, 1994).
10.91. L. HELLER and M. BEN-YAIR, Effect of Dead Sea water on Portland
cement, Journal of Applied Chemistry, No. 12, pp. 481–5 (1962).
10.92. M. BEN BASSAT, P. J. NIXON and J. HARDCASTLE, The effect of differences in the
composition of Portland cement on the properties of hardened concrete, Mag.
Concr. Res., 42, No. 151, pp. 59–66 (1990).
10.93. ACI 515.1R-79 Revised 1985, A guide to the use of waterproofing,
dampproofing, protective, and decorative barrier systems for concrete, ACI
Manual of Concrete Practice, Part 5: Masonry, Precast Concrete, Special
Processes, 44 pp. (Detroit, Michigan, 1994).
10.94. ACI 223-93, Standard practice for the use of shrinkage-compensating
concrete, ACI Manual of Concrete Practice, Part 1: Materials and General
Properties of Concrete, 29 pp. (Detroit, Michigan, 1994).
10.95. U. SCHNEIDER et al., Stress corrosion of cementitious materials in sulphate
solutions. Materials and Structures, 23, No. 134, pp. 110–15 (1990).
10.96. CONCRETE SOCIETY WORKING PARTY, Permeability Testing of Site Concrete – A
Review of Methods and Experience, Technical Report No. 31, 95 pp.
(London, The Concrete Society, 1987).

579
 10.97. D. M. ROY et al., Concrete microstructure and its relationships to pore
structure, permeability, and general durability, in Durability of Concrete, G.
M. Idorn International Symposium, ACI SP-131, pp. 137–49 (Detroit,
Michigan, 1992).
10.98. C. HALL, Water sorptivity of mortars and concretes: a review, Mag. Concr.
Res., 41, No. 147, pp. 51–61 (1989).
10.99. W. H. DUDA, Cement-Data-Book, 2, 456 pp. (Berlin, Verlag GmbH, 1984).
10.100. G. PICKETT, Effect of gypsum content and other factors on shrinkage of
concrete prisms, J. Amer. Concr. Inst., 44, pp. 149–75 (Oct. 1947).
10.101. H. H. STEINOUR, Some effects of carbon dioxide on mortars and concrete –
discussion, J. Amer. Concr. Inst., 55, pp. 905–7 (Feb. 1959).
10.102. G. J. VERBECK, Carbonation of hydrated portland cement, ASTM. Sp. Tech.
Publ. No. 205, pp. 17–36 (1958).
10.103. J. J. SHIDELER, Investigation of the moisture-volume stability of concrete
masonry units, Portl. Cem. Assoc. Development Bull, D.3 (March 1955).
10.104. I. LEBER and F. A. BLAKEY, Some effects of carbon dioxide on mortars and
concrete, J. Amer. Concr. Inst., 53, pp. 295–308 (Sept. 1956).
10.105. E. E. REIS, J. D. MOZER, A. C. BIANCHINI and C. E. KESLER, Causes and control
of cracking in concrete reinforced with high-strength steel bars – a review of
research, University of Illinois Engineering Experiment Station Bull. No.
479 (1965).
10.106. T. C. HANSEN, Cracking and fracture of concrete and cement paste, Symp. on
Causes, Mechanism, and Control of Cracking in Concrete, ACI SP-20, pp.
5–28 (Detroit, Michigan, 1968).
10.107. P. SCHUBERT and K. WESCHE, Einfluss der Karbonatisierung auf die
Eigenshaften von Zementmörteln, Research Report No. F16, 28 pp. (Institut
für Bauforschung BWTH Aachen, Nov. 1974).
10.108. A. MEYER, Investigations on the carbonation of concrete, Proc. 5th Int. Symp.
on the Chemistry of Cement, Tokyo, Vol. 3, pp. 394–401 (1968).
10.109. A. S. EL-DIEB and R. D. HOOTON, A high pressure triaxial cell with improved
measurement sensitivity for saturated water permeability of high
performance concrete, Cement and Concrete Research, 24, No. 5, pp. 854–
62 (1994).
10.110. K. KOBAYASHI, K. SUZUKI and Y. UNO, Carbonation of concrete structures and
decomposition of C-S-H, Cement and Concrete Research, 24, No. 1, pp. 55–
62 (1994).
10.111. CANADIAN STANDARDS ASSN, A23.1-94, Concrete Materials and Methods of
Concrete Construction, 14 pp. (Toronto, Canada, 1994).
10.112. BRITISH CEMENT ASSOCIATION WORKING PARTY REPORT, The Diagnosis of Alkali–
Silica Reaction, 2nd Edn, Publication 45.042, 44 pp. (Slough, BCA, 1992).
10.113. M. M. ALASALI, V. M. MALHOTRA and J. A. SOLES, Performance of various test
methods for assessing the potential alkali reactivity of some Canadian
aggregates, ACI Materials Journal, 88, No. 6, pp. 613–19 (1991).
10.114. M. G. ALI and RASHEEDUZZAFAR, Cathodic protection current accelerates
alkali–silica reaction. ACI Materials Journal, 90, No. 3, pp. 247–52 (1993).

580
 10.115. J. G. M. WOOD and R. A. JOHNSON, The appraisal and maintenance of
structures with alkali–silica reaction, The Structural Engineer, 71, No. 2, pp.
19–23 (1993).
10.116. H. WANG and J. E. GILLOTT, Competitive nature of alkali–silica fume and
alkali–aggregate (silica) reaction. Mag. Concr. Res., 44, No. 161, pp. 235–9
(1992).
10.117. M. M. ALASALI and V. M. MALHOTRA, Role of concrete incorporating high
volumes of fly ash in controlling expansion due to alkali–aggregate
reaction, ACI Materials Journal, 88, No. 2, pp. 159–63 (1991).
10.118. Z. XU, P. GU and J. J. BEAUDOIN, Application of A.C. impedance techniques in
studies of porous cementitious materials. Cement and Concrete Research, 23,
No. 4, pp. 853–62 (1993).
10.119. S. CHATTERJI, N. THAULOW and A. D. JENSEN, Studies of alkali–silica reaction.
Part 6. Practical implications of a proposed reaction mechanism, Cement and
Concrete Research, 18, No. 3, pp. 363–6 (1988).
10.120. H. CHEN, J. A. SOLES and V. M. MALHOTRA, CANMET investigations of
supplementary cementing materials for reducing alkali–aggregate
reactions, International Workshop on Alkali–Aggregate Reactions in
Concrete, Halifax, N.S., 20 pp. (Ottawa, CANMET, 1990).
10.121. D. C. STARK, Lithium admixtures – an alternative method to prevent
expansive alkali–silica reactivity. Proc. 9th International Conference on
Alkali–Aggregate Reaction in Concrete, London, Vol. 2, pp. 1017–21 (The
Concrete Society, 1992).
10.122. W. M. M. HEIJNEN, Alkali–aggregate reactions in The Netherlands, Proc. 9th
International Conference on Alkali–Aggregate Reaction in Concrete,
London, Vol. 1, pp. 432–7 (The Concrete Society, 1992).
10.123. D. LUDIRDJA, R. L. BERGER and J. F. YOUNG, Simple method for measuring
water permeability of concrete, ACI Materials Journal, 86, No. 5, pp. 433–9
(1989).
10.124. H.-J. WIERIG, Longtime studies on the carbonation of concrete under normal
outdoor exposure, RILEM Symposium on Durability of Concrete under
Normal Outdoor Exposure, Hanover, pp. 182–96 (March 1984).
10.125. BULLETIN DU CIMENT, Détermination rapide de la carbonatation du
béton, Service de Recherches et Conseils Techniques de l’Industrie Suisse du
Ciment, 56, No. 8, 8 pp. (Wildegg, Switzerland, 1988).
10.126. M. D. COHEN and A. BENTUR, Durability of portland cement–silica fume
pastes in magnesium sulfate and sodium sulfate solutions, ACI Materials
Journal, 85, No. 3, pp. 148–57 (1988).
10.127. STUVO, Concrete in Hot Countries, Report of STUVO, Dutch member
group of FIP, 68 pp. (The Netherlands, 1986).
10.128. D. W. HOBBS, Alkali–Silica Reaction in Concrete, 183 pp. (London, Thomas
Telford, 1988).
10.129. J. BIJEN, Advantages in the use of portland blastfurnace slag cement concrete
in marine environment in hot countries, in Technology of Concrete when
Pozzolans, Slags and Chemical Admixtures are Used, Int. Symp., University
of Nuevo León, pp. 483–599 (Monterrey, Mexico, March 1985).

581
 10.130. V. G. PAPADAKIS, C. G. VAYENAS and M. N. FARDIS, Physical and chemical
characteristics affecting the durability of concrete, ACI Materials
Journal, 88, No. 2, pp. 186–96 (1991).
10.131. DIN 1048, Testing of hardened concrete specimens prepared in
moulds, Deutsche Normen, Part 5 (1991).
10.132. P. B. BAMFORTH, The relationship between permeability coefficients for
concrete obtained using liquid and gas, Mag. Concr. Res., 39, No. 138, pp.
3–11 (1987).
10.133. J. D. MATTHEWS, Carbonation of ten-year concretes with and without
pulverised-fuel ash, in Proc. ASHTECH Conf., 12 pp. (London, Sept. 1984).
10.134. G. A. KHOURY, Effect of Bacterial Activity on North Sea Concrete, 126 pp.
(London, Health and Safety Executive, 1994).
10.135. BUILDING RESEARCH ESTABLISHMENT, Sulfate and acid resistance of concrete in
the ground, Digest, No. 363, 12 pp. (London, HMSO, January 1996).
10.136. J. BARON and J.-P. OLLIVIER, Eds, La Durabilité des Bétons, 456 pp. (Presse
Nationale des Ponts et Chaussées, 1992).
10.137. P. SCHUBERT and Y. EFES, The carbonation of mortar and concrete made with
jet cement, Proc. RILEM Int. Symp. on Carbonation of Concrete, Wexham
Springs, April 1976, 2 pp. (Paris, 1976).
10.138. M. SAKUTA et al., Measures to restrain rate of carbonation in concrete,
in Concrete Durability, Vol. 2, ACI SP-100, pp. 1963–77 (Detroit, Michigan,
1987).
10.139. J. BENSTED, Scientific background to thaumasite formation in concrete, World
Cement Research, Nov. pp. 102–105 (1998).
10.140. A. NEVILLE, Can we determine the age of cracks by measuring
carbonation? Concrete International, 25, No. 12, pp. 76–79 (2003) and 26,
No. 1, pp 88–91 (2004).
10.141. A. NEVILLE, Background to minimising alkali–silica reaction in concrete, The
Structural Engineer, pp. 18–19 (20 September 2005).
10.142. D. S. LANE, R. L. DETWILER and R. D. HOOTON, Testing transport properties in
concrete, Concrete International, 32, No. 11, pp. 33–38 (2010).
10.143. H. F. W. TAYLOR, C. FAMY and K. L. SCRIVENER, Delayed ettringite
formation, Cement and Concrete Research, 31, pp. 683–93 (2001).
10.144. W. G. HIME, Delayed ettringite formation. PCI Journal, 41, No. 4, pp. 26–30
(1996).

582
Chapter 11. Effects of freezing and thawing and of chlorides

This chapter is concerned with two, sometimes separate and sometimes linked,
mechanisms of damage to concrete. The first of these, although relevant only in cold
climates, is a major cause of a lack of durability of concrete unless proper precautions
are taken. The second mechanism, that is, the action of chlorides, is relevant only to
reinforced concrete but it, too, can result in extensive damage of structures. The action
of chlorides is encountered both in cold climates and in hot climates, but the details
under the two conditions differ from one another.
Action of frost
In Chapter 8, the effects of frost on fresh concrete were considered and methods of
avoiding freezing of fresh concrete were discussed. What cannot be avoided, however,
is the exposure of mature concrete to alternating freezing and thawing – a temperature
cycle frequently met with in nature.
As the temperature of saturated concrete in service is lowered, the water held in the
capillary pores in the hardened cement paste freezes in a manner similar to the
freezing in the pores in rock, and expansion of the concrete takes place. If subsequent
thawing is followed by re-freezing, further expansion takes place, so that repeated
cycles of freezing and thawing have a cumulative effect. The action takes place
mainly in the hardened cement paste; the larger voids in concrete, arising from
incomplete compaction, are usually air-filled and, therefore, not appreciably subject to
the action of frost.11.4
Freezing is a gradual process, partly because of the rate of heat transfer through
concrete, partly because of a progressive increase in the concentration of dissolved
salts in the still unfrozen pore water (which depresses the freezing point), and partly
because the freezing point varies with the size of the pore. Because the surface tension
of the bodies of ice in the capillary pores puts them under pressure that is higher the
smaller the body, freezing starts in the largest pores and gradually extends to smaller
ones. Gel pores are too small to permit the formation of nuclei of ice at temperatures
higher than –78 °C, so that in practice no ice is formed in them.11.4 However, with a
fall in temperature, because of the difference in entropy of gel water and ice, the gel
water acquires an energy potential enabling it to move into the capillary pores
containing ice. The diffusion of gel water which takes place leads to a growth of the
ice body and to expansion.11.4
There are thus two possible sources of dilating pressure. First, freezing of water
results in an increase in volume of approximately 9 per cent, so that the excess water
in the cavity is expelled. The rate of freezing will determine the velocity with which
water displaced by the advancing ice front must flow out, and the hydraulic pressure
developed will depend on the resistance to flow, i.e. on the length of path and the
permeability of the hardened cement paste in the area between the freezing cavity and
a void that can accommodate the excess water.11.5
The second dilating force in concrete is caused by diffusion of water leading to a
growth of a relatively small number of bodies of ice. Although the action of freezing
and thawing upon concrete is still debated, the latter mechanism is believed to be
particularly important in causing damage of concrete.11.6 This diffusion is caused by
osmotic pressure brought about by local increases in solute concentration due to the

583
separation of frozen (pure) water from the pore water. A slab freezing from the top
will be seriously damaged if water has access from the bottom and can travel through
the thickness of the slab due to osmotic pressure. The total moisture content of the
concrete will then become greater than before freezing, and in a few cases damage by
segregation of ice crystals into layers has actually been observed.11.7,11.47
Osmotic pressure arises also in another connection. When salts are used for de-
icing road or bridge surfaces, some of these salts become absorbed by the upper part
of the concrete. This produces a high osmotic pressure, with a consequent movement
of water toward the coldest zone where freezing takes place. The action of de-icing
salts is considered in a later section of this chapter.
When the dilating pressure in the concrete exceeds its tensile strength, damage
occurs. The extent of the damage varies from surface scaling to complete
disintegration as ice is formed, starting at the exposed surface of the concrete and
progressing through its depth. Under the conditions prevailing in a temperate climate,
road kerbs (curbs) (which remain wet for long periods) are more vulnerable to frost
than any other concrete. The second most severe conditions are those in a road slab,
particularly when salt is used for de-icing. In countries with a colder climate, the
damage due to frost is more general and, unless suitable precautions are taken, more
serious.
At this stage, it may be useful to consider why it is alternating freezing and
thawing that causes progressive damage. Each cycle of freezing causes a migration of
water to locations where it can freeze. These locations include fine cracks which
become enlarged by the pressure of the ice and remain enlarged during thawing when
they become filled with water. Subsequent freezing repeats the development of
pressure and its consequences.
While the resistance of concrete to freezing and thawing depends on its various
properties (e.g. strength of the hardened cement paste, extensibility, and creep), the
main factors are the degree of saturation and the pore system of the hardened cement
paste. The general influence of saturation of concrete is shown in Fig. 11.1: below
some critical value of saturation, concrete is highly resistant to frost,11.2 and dry
concrete is totally unaffected. In other words, if concrete is never going to be
saturated, there is no danger of damage from freezing and thawing. It may be noted
that, even in a water-cured specimen, not all residual space is water-filled, and indeed
this is why such a specimen does not fail on first freezing.11.8 A large proportion of
concrete in service dries partially, at least at some time in its life and, on rewetting,
such concrete will not re-absorb as much water as it has lost.11.9 It is desirable,
therefore, to allow concrete to dry out before exposure to winter conditions, and
failure to do so will increase the severity of frost damage. An example of the
influence of age at which first freezing takes place upon damage to concrete is shown
in Fig. 11.2.11.3

584
 Fig. 11.1. Influence of saturation of concrete on its resistance to frost expressed
by an arbitrary coefficient11.2

Fig. 11.2. Increase in volume of concrete subjected to freezing and thawing as a

function of age at which first freezing starts11.3
What is the critical value of saturation? A closed container with more than 91.7 per
cent of its volume occupied by water will, on freezing, become filled with ice, and
will become subjected to bursting pressure. Thus, 91.7 per cent can be considered to
be the critical saturation in a closed vessel. This is not, however, the case in a porous

585
body, where the critical saturation depends on the size of the body, on its
homogeneity, and on the rate of freezing. Space available for expelled water must be
close enough to the cavity in which ice is being formed, and this is the basis of air
entrainment: if the hardened cement paste is subdivided into sufficiently thin layers by
air bubbles, it has no critical saturation.
Air bubbles can be introduced by air entrainment, which is discussed later in this
chapter. Although air entrainment greatly enhances the resistance of concrete to
cycles of freezing and thawing, it is vital that the concrete has a low water/cement
ratio so that the volume of capillary pores is small. It is also essential that substantial
hydration takes place before exposure to freezing. Such concrete has a low
permeability and imbibes less water in wet weather.
Figure 11.3 shows the general effect of the absorption of concrete on its resistance
to freezing and thawing,11.99 and Fig. 11.4 illustrates the influence of the water/cement
ratio on the resistance to freezing and thawing of concrete moist-cured for 14 days
and then stored in air of 50 per cent relative humidity for 76 days prior to exposure to
freezing and thawing.11.11.

586
 Fig. 11.3. Relation between absorption of concrete and the number of cycles of
freezing and thawing required to cause a 2 per cent reduction in the mass of the
specimen11.99

Fig. 11.4. Influence of the water/cement ratio on the resistance to freezing and
thawing of concrete moist-cured for 14 days and then stored for 76 days at a
relative humidity of 50 per cent11.11
Adequate curing is vital to reduce the amount of freezable water in the paste; this
is illustrated in Fig. 11.5 for concrete with a water/cement ratio of 0.41. This figure
shows also that the freezing temperature decreases with age because of an increase in
the concentration of salts in the still remaining freezable water. In all cases, a small
amount of water freezes at 0 °C (32 °F), but this is probably free surface water on the
specimen. The temperatures at which freezing of capillary water starts were found to
be, approximately, –1 °C (30 °F) at 3 days, –3 °C (27 °F) at 7 days, and –5 °C (23 °F)
at 28 days.11.12

587
 Fig. 11.5. Effect of age of concrete on amount of water frozen, as a function of
temperature11.12
Whether or not a given concrete is vulnerable to frost, be it due to the expansion of
the hardened cement paste or of the aggregate, can be determined by cooling the
specimen through the freezing range and measuring the change in volume: frost-
resistant concrete will contract when water is transferred by osmosis from the
hardened cement paste to the air bubbles, but vulnerable concrete will dilate, as
shown in Fig. 11.6. This one-cycle test is very useful.11.23 It has been found that the
maximum dilation on first freezing correlates linearly with the residual expansion on
subsequent thawing; the latter can, therefore, also be used as an indicator of the
vulnerability of concrete.11.26

588
 Fig. 11.6. Change in volume of frost-resistant and vulnerable concretes on
cooling11.4
ASTM C 671-94 prescribed a test method for the critical dilation of concrete
subjected to repeated two-week cycles of short freezing and prolonged storage in
water. The length of time until critical dilation occurs can be used to rank concretes in
terms of their resistance to freezing and thawing under the given conditions. This
standard has been withdrawn.
Behaviour of coarse aggregate particles
Consideration of critical saturation applies also to individual particles of coarse
aggregate. An aggregate particle by itself will not be vulnerable if it has a very low
porosity, or if its capillary system is interrupted by a sufficient number of macropores.
However, an aggregate particle in concrete can be considered as a closed container,
because the low permeability of the surrounding hardened cement paste will not allow
water to move sufficiently rapidly into air voids. Thus an aggregate particle saturated
above 91.7 per cent will, on freezing, destroy the surrounding mortar.11.4 It may be
recalled that common aggregates have a porosity of 0 to 5 per cent, and it is preferable
to avoid aggregates of high porosity. However, the use of such aggregates need not
necessarily result in frost damage. Indeed, large pores present in aerated concrete and
in no-fines concrete probably contribute to the frost resistance of those materials.

589
Furthermore, even with ordinary aggregate, no simple relation between the porosity of
the aggregate and the resistance to freezing and thawing of the concrete has been
established.
If a vulnerable particle is near the surface of the concrete, instead of disrupting the
surrounding hardened cement paste, it can cause a popout.
The effect of drying of aggregate, prior to mixing, on the durability of concrete is
shown in Fig. 11.7. It can be seen that the presence of saturated aggregate, particularly
of large size, can result in the destruction of concrete, whether or not the concrete is
air entrained. On the other hand, if the aggregate is not saturated at the time of mixing,
or if it is allowed to dry partially after placing and the capillaries in the paste are
discontinuous, re-saturation is not easily achieved except during a prolonged period of
cold weather.11.1 On rewetting of concrete, it is the hardened cement paste that tends to
be more nearly saturated than the aggregate, as water can reach the aggregate only
through the paste, and also because the finer-textured paste has a greater capillary
attraction. As a result, the hardened cement paste is more vulnerable but it can be
protected by air entrainment.

Fig. 11.7. Relation between the condition of aggregate before mixing and the
number of cycles of freezing and thawing to produce a 25 per cent loss in the
mass of the specimen11.10
Air entrainment of the cement paste does not alleviate the effects of freezing of
coarse aggregate particles.11.92 Nevertheless, aggregate should be tested in air-entrained
concrete in order to exclude the effect of the durability of the surrounding hardened
cement paste. For this reason, ASTM C 682-94 (withdrawn) provides for the
evaluation of the frost resistance of coarse aggregate when used in air-entrained
concrete using the test for the critical dilation of concrete subjected to freezing given
in ASTM C 671-94 (withdrawn).
A test for frost heave of unbound aggregate is prescribed in BS 812 : 124 : 2009;
although not directly applicable to aggregate in concrete, the test may be of interest in
a preliminary investigation of previously unused aggregates.
There is one type of cracking of concrete road, bridge, and airfield surfaces which
is particularly linked to aggregate. This is called D-cracking. It consists of the
development of fine cracks near free edges of slabs, but the initial cracking starts

590
lower in the slab where moisture accumulates and the coarse aggregate becomes
saturated to the critical level. Thus, we have essentially a failure of aggregate, which,
with cyclic freezing and thawing, becomes slowly saturated and causes failure of the
surrounding mortar.11.25 D-cracking can manifest itself very slowly, sometimes
reaching the top of the slab only after 10 or 15 years, so that assignment of
responsibility for failure is difficult.
Aggregates associated with D-cracking are nearly always of sedimentary origin
and can be calcareous or siliceous. They can be gravel or crushed rock. While the
absorption characteristics of the aggregate are clearly relevant to the proneness of
concrete to D-cracking, the absorption value alone does not distinguish durable
aggregate from non-durable. Freezing and thawing laboratory tests on concrete
containing the given aggregate give a good indication of likely behaviour in service. If
after 350 cycles the expansion is less than 0.035 per cent, D-cracking will not
develop.11.25 It should be noted that the same parent rock leads to less D-cracking when
the aggregate particles are smaller (see Fig. 11.8); thus, comminution of a given
aggregate may reduce the risk of D-cracking.11.25

591
 Fig. 11.8. Relation between maximum aggregate particle size and expansion in
laboratory freezing and thawing tests. A failure criterion of 0.035 per cent
expansion in 350 or fewer cycles is indicated11.25
More generally, large aggregate particles are more vulnerable to
frost.11.34 Furthermore, the use of aggregate with a large maximum size or a large
proportion of flat particles is inadvisable as pockets of bleed water may collect on the
underside of the coarse aggregate. It is relevant to note that air entrainment reduces
bleeding.
Air entrainment
Because the damaging action of freezing and thawing involves expansion of water on
freezing, it is logical to expect that, if excess water can readily escape into
adjacent air-filled voids, damage of concrete will not occur. This is the underlying
principle of air entrainment. It should be emphasized, however, that the volume of
capillary pores should be minimized in the first place as otherwise the volume of
freezable water would exceed that which can be accommodated by the deliberately
entrained air voids. This requirement translates into the need for an adequately low
water/cement ratio, which also ensures a strength of concrete such that it can better
resist the damaging forces induced by freezing. According to ACI 201.2R11.92 to be
resistant to freezing and thawing, concrete should have a water/cement ratio not
greater than 0.50; this is reduced to 0.45 in thin sections including bridge decks and
kerbs (curbs). Alternatively, concrete should not be exposed to cycles of freezing and
thawing until its strength has reached 24 MPa (3500 psi).
Entrained air in concrete is defined as air intentionally incorporated by means of a
suitable agent. This air should be clearly distinguished from accidentally entrapped air:
the two kinds differ in the magnitude of the air bubbles, those of entrained air having
typically a diameter of about 50 μm (0.002 in.), whereas accidental air usually forms
very much larger bubbles, some as large as the familiar, albeit undesirable, pockmarks
on the formed surface of concrete.
Entrained air produces discrete, nearly spherical, bubbles in the cement paste so
that no channels for the flow of water are formed and the permeability of the concrete
is not increased. The voids never become filled with the products of hydration of
cement as gel can form only in water.
The improved resistance of air-entrained concrete to frost attack was discovered
accidentally when cement ground with beef tallow, added as a grinding aid, was
observed to make more durable concrete than when no grinding aid was used. The
main types of air-entraining agents are:
(a) salts of fatty acids derived from animal and vegetable fats and oils (beef tallow
being an example of this group)
(b) alkali salts of wood resins, and
(c) alkali salts of sulfated and sulfonated organic compounds.
All these agents are surface-active agents, or surfactants, that is, long-chain
molecules which orient themselves so as to reduce the surface tension of the water,
the other end of the molecule being directed toward the air. Thus, the air bubbles
formed during mixing become stabilized: they are covered by a sheath of air-
entraining molecules repelling one another, and so preventing coalescence and
ensuring a uniform dispersion of the entrained air.

592
 Numerous types of air-entrained agents are available in the form of commercial
admixtures, but the performance of unknown ones should be checked by trial mixes.
ASTM C 260-06 and BS EN 934-2320 as well as BS EN 934-6 : 2001, lay down the
performance requirements of air-entraining agents, usually called admixtures. The
essential requirements of an air-entraining admixture are that it rapidly produces a
system of finely divided and stable foam, the individual bubbles of which resist
coalescence. The foam must have no harmful chemical effect on the cement.
The air-entraining admixture is normally dispensed into the mixer direct in the
form of a solution. The timing of the discharge of the admixture into the mixer is of
importance so as to ensure a uniform distribution and adequate mixing for the
formation of the foam. If other admixtures are also used, they should not come into
contact with the air-entraining admixture prior to entering the mixer because their
interaction could affect their performance.
Air-entraining agents can also be interground with cement but this allows no
flexibility in the air content of the concrete, so that the use of air-entrained cements
should generally be limited to minor construction.
Air-void system characteristics
Because the resistance to the movement of water through hardened cement paste must
not be excessive to the point of preventing the flow, it follows that water, wherever it
is located, must be sufficiently close to air-filled space, that is to the bubbles of
entrained air. Thus, the fundamental requirement which ensures the efficacy of air
entrainment is a limit on the maximum distance which the escaping water has to travel.
The practical factor is the spacing of the air bubbles, i.e. the thickness of the hardened
cement paste between adjacent air voids, which is twice the maximum distance
referred to above. Powers11.15 calculated that an average spacing of 250 μm (0.01 in.)
between the voids is required for full protection from frost damage (Fig. 11.9);
nowadays, 200 μm (0.008 in.) is usually recommended.11.94

593
Fig. 11.9. Relation between durability and spacing of bubbles of entrained air11.16
Because the total volume of voids in a given volume of concrete affects the
strength of concrete (cf. p. 280) it follows that, for a given spacing, the air bubbles
should be as small as possible. Their size depends to a large degree on the foaming
process used. In fact, bubbles are not all of one size, and it is convenient to express
their size in terms of specific surface (square millimetres per cubic millimetre or
square inches per cubic inch).
It should not be forgotten that accidental (entrapped) air is present in any concrete,
whether air-entrained or not, and, as the two kinds of voids cannot be distinguished
other than by direct observation, the specific surface represents an average value for
all voids in a given cement paste. For air-entrained concrete of satisfactory quality,
the specific surface of voids is in the range of approximately 16 to 24 mm–1 (400 to
600 in.–1), but sometimes it is as high as 32 mm–1 (800 in.–1). By contrast, the specific
surface of accidental air is less than 12 mm–1 (300 in.–1).11.15
The adequacy of air entrainment in a given hardened concrete can be estimated by
a spacing factor, , determined by a test method prescribed in ASTM C 457-10a. The
spacing factor is a useful index of the maximum distance of any point in the hardened

594
cement paste from the periphery of a nearby air void. The calculation of the factor is
based on the assumption that all air voids are equal-sized spheres arranged in a simple
cubic lattice. The calculation is laid down by ASTM C 457-10a and requires the
knowledge of: the air content of the concrete, using a linear traverse microscope to
determine the average number of air void sections per inch or the average chord
intercept of the voids; and the hardened cement paste content by volume. The spacing
factor is expressed in inches or millimetres; usually a value of not more than 200 μm
(0.008 in.) is a maximum value required for satisfactory protection from freezing and
thawing.
It may be useful to add that the water which has moved into air voids during
freezing returns into the smaller capillary pores in the hardened cement paste during
thawing. Thus, protection by air entrainment continues permanently for repeated
freezing and thawing.11.17 Rapid thawing followed by freezing is not harmful, as the
water is already in the air voids; on the other hand, slow thawing followed by very
rapid freezing may not allow sufficient movement of water to take place.
Entrained-air requirements
From the requirement of a maximum spacing of air voids, it is possible to calculate
the minimum volume of entrained air in the hardened cement paste. For each mix,
there is a minimum volume of voids required. Klieger11.14 found this volume to
correspond to 9 per cent of the volume of mortar. As the volume of hardened cement
paste, in which alone the air is entrained, varies with the richness of the mix, the air
content of concrete which is required depends on the mix proportions; in practice, the
maximum size of aggregate is used as a parameter.
For a given air content, the spacing of air voids depends on the water/cement ratio
of the mix as shown in Fig. 11.10. Specifically, the higher the water/cement ratio the
larger the bubble spacing (and the lower the specific surface) because small bubbles
coalesce.11.42 The stability of air bubbles is considered on p. 553.

595
 Fig. 11.10. Influence of the water/cement ratio on the void spacing in concrete
with an average air content of 5 per cent11.11
Typical values of the amount of air required for 250 μm (0.01 in.) spacing for
different mixes are given in Table 11.1, based on Powers’ results.11.15 A higher specific
surface, which corresponds to smaller bubbles, is desirable so as to minimize the
adverse effect of the air in concrete on its strength. Table 11.1 indicates that, for a
particular value of the specific surface of the air voids, richer mixes require a greater
volume of entrained air than lean ones. However, the richer the mix the greater the
specific surface of the voids for a given air content. This is illustrated in Table 11.2,
based on ref. 11.14.

Table 11.1. Air Content Required for a Void Spacing of 250 μm (0.01 in.)11.15

596
Table 11.2. Example of the Influence of the Cement Content of the Mix upon the
Specific Surface of Air Voids in Concrete with a Maximum Size of Aggregate of
19 mm ( in.) (based on ref. 11.14)

It can be noted that appropriately higher values may be required in grout in

prestressed concrete ducts; the voids induced by aluminium powder which reacts with
the alkalis, used to ensure complete filling of a duct, are insufficient for frost
protection.
The severity of exposure of the concrete affects the value of the air content which
should be specified,11.92 as shown in Table 11.3 in which “severe exposure” describes
conditions such that concrete may be in almost continuous contact with moisture prior
to freezing or where de-icing salts are used; the air content in mortar is expected to be
9 per cent. “Moderate exposure” describes conditions when concrete is only
occasionally exposed to moisture prior to freezing and when no de-icing salts are used;
the air content in mortar is expected to be 7 per cent. A tolerance of ± per cent is

597
permitted on the values given in Table 11.3. Table 11.3 includes also the British
requirements; these are less demanding than specified by ACI 201.2R-92.11.92 On the
other hand, Swedish requirements are similar to those of ACI 201.2R-92 but the
tolerance permitted under very aggressive conditions is only ±1 per cent.11.43

Table 11.3. Recommended Air Content of Concretes Containing Aggregates of

Different Maximum Size

Some standards specifiy not only a maximum value of the bubble spacing but also
a minimum value of the specific surface of the air in concrete so as to ensure the
presence of small air bubbles. This gives the best protection from freezing and
thawing, coupled with the least loss of strength due to the presence of voids in the
concrete.
Factors influencing air entrainment
The volume of air entrained in a given concrete is independent of the volume of
entrapped air and depends primarily on the amount of air-entraining admixture added.
The larger the quantity of the admixture the more air is entrained, but there is a
maximum amount of any admixture beyond which there is no increase in the volume
of voids.
To obtain a desired percentage of entrained air in concrete, for any given air-
entraining admixture, there is a recommended dosage. However, the actual amount of
air which becomes entrained is affected by a number of factors. Broadly speaking, for
a given percentage of entrained air, more admixture is required under the following
conditions:
when the cement has a higher fineness;
when the cement has a low alkali content;
when fly ash is incorporated in the mix, the more so the higher the carbon
content in the fly ash;

598
 when the aggregate has a high proportion of ultrafine material or when finely-
divided pigments are used;
when the concrete temperature is high;
when the workability of the mix is low; and
when the mixing water is hard.
In connection with water, it can be mentioned that water used to wash truck mixers
is very hard, especially if the mix used was air-entrained; the difficulty of entraining
air is alleviated if the air-entraining admixture is not added with the wash water but
with the additional clean water or with the sand.11.95
Mixes with high cement contents, about 500 kg/m3 (840 lb/yd3), and very low
water/cement ratios (0.30 to 0.32), which are used in low-slump concrete overlays for
bridge decks, require extremely high admixture dosages.11.48
Air entrainment can be used with various types of cement. However, there may be
difficulties with mixes containing fly ash. The main reason for this is that carbon in
the fly ash, arising from imperfect combustion, can absorb the surface-active air-
entraining agent, thereby reducing its effectiveness.11.38 In consequence, an increased
dosage of the air-entraining admixture may be required but, if the active carbon
content is not uniform, a variable air content may result.
In addition, it has sometimes been observed that properly entrained air can become
destabilized in the presence of carbon particles in the fly ash. Thus, the air content of
the mix decreases prior to placing. This may be due to the adsorption of the air
bubbles on the highly active surface of the carbon particles.11.38 Special air-entraining
admixtures which contain a polar species, preferentially adsorbed by carbon, have
been developed but they cannot remedy the difficulties unless there is no variation in
the nature of the carbon.11.38
Air entrainment can be used when silica fume is incorporated in the mix; resistance
to freezing and thawing is ensured by the usual spacing factor not larger than
200 μm.11.35
Air-entraining admixtures can be used when other admixtures are also included in
the mix. When water-reducing admixtures are used at the same time as air-entraining
admixtures, a lesser amount of the latter is frequently needed for a given percentage
of air, even if the water-reducing admixture has no air-entraining properties per se.
The explanation is that the physical or chemical environment is altered so as to permit
the air-entraining admixture to operate more efficiently.11.27 It should be noted that
combinations of some admixtures may be incompatible, so that tests with the actual
materials to be used should always be made. Indeed, trial mixes to determine the
required dosage of any given air-entraining admixture are highly recommended.
Some superplasticizers, in combination with certain cements and air-entraining
admixtures, may produce an unstable void system; it is, therefore, vital to check their
compatibility.11.44 Given such compatibility, satisfactory air entrainment of concrete
containing a superplasticizer is possible, but there is usually a slight increase in the
bubble size with a consequent increase in the bubble spacing factor.11.52 For this reason,
some increase in the dosage of the air-entraining admixture is
required.11.51 Nevertheless, at water/cement ratios below 0.4, concretes containing a
superplasticizer exhibit good resistance to freezing and thawing when the spacing
factor is somewhat larger than normally required, namely up to 240 μm.11.100 Indeed,
Canadian standards allow a maximum spacing factor of 230 μm.

599
 The actual mixing operation also affects the resultant air content, and the loading
sequence can have a significant effect. The cement should be well dispersed and the
mix uniform before the air-entraining admixture is introduced.11.46 If the mixing time is
too short, the air-entraining admixture does not become sufficiently dispersed, but
over-mixing gradually expels some air, so that there is an optimum value of mixing
time. In practice, the mixing time is fixed from other considerations, usually at a value
shorter than the minimum necessary for the admixture to become fully dispersed, and
the amount of the air-entraining admixture must be adjusted accordingly. A very fast
rotation of the mixer increases the amount of entrained air. Agitating up to 300
revolutions appears to lead only to a small loss of air (see Fig. 11.11)11.28 but after 2
hours a loss of up to 20 per cent of the original air content can occur.11.33 In some cases,
a loss as high as 50 per cent was reported.11.50

Fig. 11.11. Relation between air content and number of revolutions of the mixer.
Batches of 6 m3 (8 yd3) were mixed at 18 rpm and agitated at 4 rpm11.28
Excessive finishing operations can result in a loss of entrained air from the surface
zone of concrete, and it is this zone that is particularly vulnerable to freezing and
thawing, as well as to the action of de-icing agents.
Stability of entrained air
Assuring an adequate percentage of air in fresh concrete is not sufficient: the air voids
must be stable so that they remain in position when the concrete hardens. Indeed,
what is crucial is not the total air content but the spacing of the small air bubbles.
Three mechanisms of instability may operate.11.42 In the first, during transporting
and compacting of concrete, large bubbles move upwards by buoyancy (and also
toward the side formwork) and are lost. This has little effect on the resistance to

600
freezing and thawing, and can be even beneficial in that the loss of strength of
concrete occasioned by the inclusion of voids is reduced.
The second mechanism involves the collapse of bubbles by pressure (arising from
surface tension) which is largest in the smallest bubbles; the air becomes dissolved in
the pore water. The loss of these bubbles has a detrimental effect on the resistance of
concrete to freezing and thawing. It is likely that this mechanism of loss of the
smallest bubbles is unavoidable and it explains the frequent absence of bubbles
smaller than about 10 μm.11.42
The third mechanism consists of coalescence of small bubbles with larger ones,
also in consequence of the relation between the solubility of air and the bubble size;
the physics of this mechanism is rather complex.11.42 The formation of larger bubbles,
and therefore increased bubble spacing, is detrimental to resistance of concrete to
freezing and thawing. Moreover, because the pressure in a larger bubble is smaller
than in the original small bubble, the total volume of the coalesced bubble is larger.
This can explain why, on occasion, the volume of entrained air in hardened concrete
is higher than it was in the fresh concrete.11.42 The increased total volume of air has a
negative effect on the strength of concrete.
As far as the influence of cement on stability is concerned, it seems that stability
increases with an increase in the content of alkalis in the cement.11.45 Silica fume, at
least up to a 10 per cent content in the blended cement, does not affect the stability of
the air-void system.11.57
In practice, loss of air occurs in transporting and during vibration of concrete: the
loss is generally less than 1 percentage point, but slightly more in high-workability
concrete. For the most part, it is the larger bubbles that are expelled, so that the effect
on the resistance of concrete to freezing and thawing is small. Under normal
conditions of pumping, the loss of air is between 1 and 1.5 percentage
points.11.54 However, a much larger loss can occur during pumping when a boom is
used in a vertical position, so that the concrete in the pipeline can slide down under
gravity: the air bubbles then expand but fail to re-form when the concrete leaves the
pipeline. A remedy lies in providing resistance before discharge by an additional
length of a horizontal flexible hose.11.54
Because of the possible loss of air, air content should be determined on concrete as
placed and not only at the point of discharge from the mixer; however, determination
at the mixer may be of value as a means of batching control.
It may be noted that steam curing of air-entrained concrete may lead to incipient
cracking because of the expansion of air.
Air entrainment by microspheres
The main difficulty with the use of air-entraining admixtures is that the air content of
the concrete cannot be controlled directly: the quantity of the admixture is known but,
as mentioned earlier, the actual air content in the hardened concrete and the spacing of
air bubbles are affected by many factors. This difficulty is obviated if, instead of air
bubbles, rigid-foam particles of suitable size are used. Such easily compressible
hollow plastic microspheres (modelled on medication microcapsules) are
manufactured.11.29 They have a diameter of 10 to 60 μm (0.0004 to 0.002 in.), which is
a narrower range of sizes than is the case with entrained air bubbles. In consequence,
a lesser volume of microspheres can be used for the same protection from freezing
and thawing, so that the loss of strength of concrete is smaller. Using 2.8 per cent of

601
microspheres by volume of hardened cement paste gives a spacing factor of 70 μm
(0.003 in.),11.29 which is well below the value of 250 μm (0.01 in.) normally
recommended with entrained air.
The specific gravity of the microspheres is 45 kg/m3 (2.8 lb/ft3) and they improve
the workability of concrete to the same extent as entrained air, even though their total
volume in the mix is smaller; the reason is that they are all small.
The microspheres are available pre-mixed with 90 per cent water in the form of a
paste, and are stable except when concrete is over-mixed. They do not interact with
other admixtures, but failure to perform in the presence of superplasticizers has been
reported.11.53 The main drawback of microspheres is their high cost so that their use is
limited to special applications.
Use of highly porous particulate additives such as vermiculite, perlite or
pumice,11.49 although attractive when concrete is extruded or vacuum-dewatered, leads
to a high loss of strength and is limited to high water/cement ratios.
Measurement of air content
There are three methods of measuring the total air content of fresh concrete. Because
the entrained air cannot be distinguished in these tests from the large bubbles of
accidental air, it is important that the concrete tested be properly compacted.
The gravimetric method is the oldest one. It relies simply on comparing the density
of compacted concrete containing air, ρa, with the calculated density of air-free
concrete of the same mix proportions, ρ. The air content, expressed as a percentage of
the total volume of the concrete, is then 1 – ρa/ρ. This method is covered by ASTM C
138-09 and can be used when the specific gravity of the aggregate and the mix
proportions are constant. An error of 1 per cent in the calculated air content is not
uncommon; this order of error would be expected from the simple experience of
determining the density of nominally similar test specimens of non-air-entrained
concrete.
In the volumetric method the difference in the volumes of a sample of compacted
concrete before and after the air has been expelled is determined. The air is removed
by agitating, inverting, rolling, and rocking, the operation being performed in a
special two-part vessel. The details of the test are prescribed by ASTM C 173-10. The
main difficulty lies in the fact that the mass of water replacing the air is small
compared with the total mass of the concrete. The method is appropriate for concrete
containing any type of aggregate.
The most popular method, and one best suited for site use, is the pressure method.
It is based on the relation between the volume of air and the applied pressure (at a
constant temperature) given by Boyle’s law. The mix proportions or the properties of
the materials need not be known and, when commercial air meters are used, no
calculations are required as direct graduations in percentage of air are provided.
However, at high altitudes, the pressure meter must be re-calibrated. The meter is not
suitable for use with porous aggregates or with lightweight concrete.
A typical pressure-type air meter is shown in Fig. 11.12. The procedure consists
essentially of observing the decrease in the volume of a sample of compacted concrete
when subjected to a known pressure. The pressure is applied by a small pump, such as
a bicycle pump, and measured by a pressure gauge. Due to the increase in pressure
above atmospheric, the volume of air in the concrete decreases and this causes a fall
in the level of the water above the concrete. By arranging the level of the water to

602
vary within a calibrated tube, the air content can be read direct by an unskilled
operator.

Fig. 11.12. Pressure-type air meter

The test is covered by ASTM C 231-09 and BSEN 12350-7 : 2009, and provides
the most dependable and accurate method of determining the air content of concrete.
The tests should be performed at the point of placing the concrete so as to exclude
the air lost in transportation; preferably, concrete after compaction should be tested. It
should be remembered that what is measured is the total volume of air in the concrete,
and not just the entrained air with the desired air-void characteristics.
On the other hand, a detailed knowledge of the air-void system
of hardened concrete can be obtained from polished sections of concrete by means of

603
a microscope using the linear traverse technique11.19 or a modified point-count method
prescribed by ASTM C 457-10a.
Tests of resistance of concrete to freezing and thawing
There exist no standard methods for the determination of the resistance of concrete to
cycles of freezing and thawing such as may occur in service. However, ASTM C 666-
03 (2008) prescribes two procedures for the determination of the resistance of
concrete to rapidly repeated cycles of freezing and thawing; these procedures can be
used to compare various mixes. In Procedure A, both freezing and thawing take place
in water; in Procedure B, freezing takes place in air but thawing takes place in water.
Freezing saturated concrete in water is much more severe than in air,11.21 and the degree
of saturation of the specimen at the beginning of the tests also affects the rate of
deterioration. British Standard BS 5075 : 2 : 1982 also prescribes freezing in water.
The deterioration of concrete can be assessed in several ways. The most common
method is to measure the change in the dynamic modulus of elasticity of the specimen,
the reduction in the modulus after a number of cycles of freezing and thawing
expressing the deterioration of the concrete. This method indicates damage before it
has become apparent either visually or by other methods, although there are some
doubts about this interpretation of the decrease in the modulus after the first few
cycles of freezing and thawing.11.20
With the ASTM methods it is usual to continue freezing and thawing for 300
cycles or until the dynamic modulus of elasticity is reduced to 60 per cent of its
original value, whichever occurs first. The durability can then be assessed as:

There are no established criteria for acceptance or rejection of concrete in terms of the
durability factor; its value is thus primarily in a comparison of different concretes,
preferably when only one variable (e.g. aggregate) is changed. However, some
guidance in interpretation can be obtained from the following: a factor smaller than 40
means that the concrete is probably unsatisfactory with respect to resistance to
freezing and thawing; 40 to 60 is the range for concretes with doubtful performance;
above 60, the concrete is probably satisfactory; and around 100 it can be expected to
be satisfactory.
The effects of freezing and thawing can also be assessed from measurements of the
loss of compressive or flexural strength or from observations of the change in
length11.20 (used in ASTM C 666-03 (2008) and in BS 5075-2 : 1992) or in the mass of
the specimen. A large change in length is an indication of internal cracking: a value of
200 × 10–6 for tests in water is taken to represent serious damage.11.60
Measurement of a decrease in the mass of the specimen is appropriate when
damage takes place mainly at the surface of the specimen, but is not reliable in cases
of internal failure; the results depend also on the size of the specimen. It may be noted
that, if failure is primarily due to unsound aggregate, it is more rapid and more severe
than when the hardened cement paste is disrupted first. It should be added that the
tests of ASTM C 666-03 (2008) are useful in evaluating the possibility of the
development of D-cracking due to the unsoundness of coarse aggregate.11.36

604
 Another test method determined the dilation of concrete subjected to slow freezing,
and was prescribed by ASTM C 671-94 (withdrawn); this is referred to on p. 544.
It can be seen that a number of tests and of means of assessing the results are
available, and it is not surprising that the interpretation of test results is difficult. If the
tests are to yield information indicative of the behaviour of concrete in practice, the
test conditions must not be fundamentally different from the field conditions. A major
difficulty lies in the fact that a test must be accelerated in comparison with the
conditions of outdoor freezing, and it is not known at what stage acceleration affects
the significance of the test results. One difference between the conditions in the
laboratory and actual exposure lies in the fact that, in the latter case, there is seasonal
drying during the summer months but, with permanent saturation imposed in some of
the laboratory tests, all the air voids can eventually become saturated with a
consequent failure of the concrete. Indeed, probably the most important factor
influencing the resistance of concrete to cycles of freezing and thawing is the degree
of its saturation,11.58 and this may increase by prolonged accretion of ice during the
freezing period; an example of such exposure occurs in Arctic waters. The duration of
the freezing period in water is, therefore, of importance.
An important feature of the tests of ASTM C 666-03 (2008) is that cooling takes
place at a rate of up to 11 °C/h (20 °F/h) whereas, in practice, 3 °C/h (5 °F/h) is more
usual. The maximum rate of cooling of outdoor air in Europe was reported by
Fagerlund11.58 as 6 °C/h (11 °F/h). However, when radiation towards a clear sky can
occur on a winter night, the surface temperature of concrete can cool at the rate of
12 °C/h (22 °F/h) even though the ambient air cools at 6 °C/h (11 °F/h).
The influence of the rate of freezing upon the resistance of concrete to cycles of
freezing and thawing was demonstrated by Pigeon et al.;11.59 as shown in Fig. 11.13,
the higher the rate of freezing the smaller the spacing factor required for the
protection of concrete.

605
Fig. 11.13. Relation between the rate of freezing and the spacing factor required
for the protection of concrete with a water/cement ratio of 0.5. The line
represents data from ref. 11.59 and the points those from ref. 11.15
The vulnerability of concrete (with a water/cement ratio lower than 0.5) to freezing
and thawing in service depends on the degree of hydration of the cement paste: time is
required for a dense pore structure to develop. The normal procedure of ASTM C
666-03 (2008) requires testing at the age of 14 days, which may be far too early.
However, the test method provides for the choice of some other age.
It can be stated that some accelerated freezing and thawing tests result in the
destruction of concrete that in practice could be satisfactory.11.22 However, the ability of
a concrete to withstand a considerable number of laboratory freezing and thawing
cycles (say 150) is a probable indication of its high degree of durability under service
conditions. The ASTM C 666-03 (2008) tests show, however, a high scatter in the
middle range of durability. Whereas the numbers of cycles of freezing and thawing in
a test and in actual concrete are not simply related, it may be interesting to note that,
in much of the United States, there are more than 50 cycles per annum.
The number of cycles of freezing and thawing to which a particular concrete
element is exposed in service is not readily determined. The record of air temperature
is inadequate. For example, the situation is complicated on a sunny day with passing
clouds. The temperature of the surface of the concrete directly exposed to the sun can
rise by 10 °C (18 °F) above the air temperature. When the sky clouds over, the
concrete cools.11.96 Thus, several cycles of freezing and thawing can occur in a day.

606
These events are influenced by the angle of incidence of the solar radiation, so that a
south-facing exposure may be most harmful. These rapid temperature changes at the
surface of the concrete can also induce harmful temperature gradients.11.96 It can be
mentioned in passing that, in some northern locations, there is only one cycle of
freezing and thawing per annum: its duration is six months.
Further effects of air entrainment
The original purpose of air entrainment was to make concrete resistant to freezing and
thawing. This is still the most common reason for incorporating entrained air in
concrete, but there are some further effects of air entrainment on the properties of
concrete, some beneficial, others not. One of the most important is the influence of
voids on the strength of concrete at all ages. It will be remembered that the strength of
concrete is a direct function of its density ratio, and voids caused by entrained air will
affect the strength in the same way as voids of any other origin. Figure 11.14 shows
that when entrained air is added to a mix, without any other change in the mix
proportions being made, the decrease in the strength of concrete is proportional to the
volume of air present. The range considered is up to 8 per cent of air and this is why
the curved part of the strength–void ratio relation is not apparent (cf. Fig. 4.1). That
the origin of the air is irrelevant is apparent from the dotted curve in Fig. 11.14 which
shows the strength–air content relation for the case when the voids are due to
inadequate compaction, as well as when they are due to entrainment. The range of
tests covered mixes with water/cement ratios between 0.45 and 0.72, and this shows
that the loss of strength expressed as a fraction of the strength of air-free concrete is
independent of the mix proportions. The average loss of compressive strength is 5.5
per cent for each percentage point of air present.11.18 The effect on flexural strength is
much smaller. The relation between the volume of voids in concrete and loss of
strength was confirmed by Whiting et al.11.55

607
 Fig. 11.14. Effect of entrained and accidental air on the strength of concrete11.18
It should be noted that strength is affected by the total volume of all the voids
present: entrapped air, entrained air, capillary pores, and gel pores. When entrained air
is present in the concrete, the total volume of capillary pores is smaller because a part
of the gross volume of the hardened cement paste consists of entrained air. This is not
a negligible factor because the volume of entrained air represents a significant
proportion of the gross volume of the hardened cement paste. For instance, in a
1:3.4:4.2 mix with a water/cement ratio of 0.80, the capillary pores at the age 7 days
were found to occupy 13.1 per cent of the volume of concrete. With entrained air in a
mix of the same workability (1:3.0:4.2 with a water/cement ratio of 0.68), the
capillary pores occupied 10.7 per cent, but the volume of air (entrained and entrapped)
was 6.8 per cent (compared with 2.3 per cent in the former mix).11.24
This is one reason why air entrainment does not cause as large a loss of strength as
might be expected. But a more important reason is that the entrainment of air has a
considerable beneficial effect on the workability of the concrete. As a result, in order

608
to keep the workability constant, the addition of entrained air can be accompanied by
a reduction in the water/cement ratio, compared with a similar mix without entrained
air. For very lean mixes, say, with an aggregate/cement ratio of 8 or more, and
particularly when angular aggregate is used, the improvement in workability due to
air entrainment is such that the resultant decrease in the water/cement ratio
compensates fully for the loss of strength due to the presence of the voids. In the case
of massive structures, where the development of heat of hydration of cement, and not
strength, is often of primary importance, air entrainment permits the use of mixes with
low cement contents and, therefore, a low temperature rise. In richer mixes, the effect
of air entrainment on workability is smaller, so that the water/cement ratio can be
lowered only a little, and there is a net loss in strength. In general terms, entrainment
of 5 per cent of air increases the compacting factor of concrete by about 0.03 to 0.07,
and the slump by 15 to 50 mm ( to 2 in.),11.18 but actual values vary with the
properties of the mix. Air entrainment is also effective in improving the workability of
the rather harsh mixes made with lightweight aggregate.
The reason for the improvement of workability by the entrained air is probably that
the air bubbles, kept spherical by surface tension, act as a fine aggregate of very low
surface friction and considerable elasticity. Entraining air in the mix makes it actually
behave like an over-sanded mix and, for this reason, the addition of entrained air
should be accompanied by a reduction in the sand content. The latter change allows a
further reduction in the water content of the mix, i.e. a further compensation of the
loss of strength due to the presence of voids is possible.
It is interesting to note that air entrainment affects the consistency or ‘mobility’ of
the mix in a qualitative manner: the mix can be said to be more ‘plastic’, so that for
the same workability, as measured, say, by the compacting factor, the mix containing
entrained air is easier to place and compact than an air-free mix.
The presence of entrained air is also beneficial in reducing bleeding: the air
bubbles appear to keep the solid particles in suspension so that sedimentation is
reduced and water is not expelled. For this reason, permeability and the formation of
laitance are also reduced, and this results in an improved resistance to freezing and
thawing of the top layer of a slab or a lift. This is relevant to the beneficial effect of
air entrainment on the destructive action of de-icing agents. Air entrainment reduces
segregation during handling and transporting as the mix is more cohesive, but
segregation due to over-vibration is still possible, particularly as, under those
conditions, the air bubbles are expelled.
The addition of entrained air lowers the density of the concrete and makes cement
and aggregate ‘go further’. This offers an economic advantage but is offset by the cost
of the air-entraining admixture and the associated operations.
Effects of de-icing agents
Horizontal surfaces, such as road slabs and bridge decks which are subjected to
freezing and thawing, are often also treated by de-icing agents for the purpose of
removing snow and ice. These agents have an adverse effect on concrete, leading to
surface scaling and sometimes to corrosion of reinforcement. The latter topic is dealt
with later in this chapter.
The salts commonly used are NaCl and CaCl2, the latter being more expensive. The
salts produce osmotic pressure and cause movement of water toward the top layer of
the slab where freezing takes place,11.4 and hydraulic pressure is developed.11.92 Thus,

609
the action is similar to ordinary freezing and thawing, but is more severe. Indeed, the
damage caused by the de-icing agents is primarily physical, and not chemical, in
nature11.13 and is independent of whether the de-icer is organic or not, or is a salt or
not.11.31 However, there is also some possibility of leaching of Ca(OH)2 which has a
greater solubility in a chloride solution than in water;11.32 it is also possible for
chloroaluminates to form under wetting and drying.11.32
Mather11.30 suggested the following sequence. The de-icing agent melts the snow or
ice, the resulting water being often ponded by adjacent ice. The water is actually a salt
solution and, therefore, has a lowered freezing point. Some of this solution is
absorbed by the concrete which may become saturated. As more ice melts, the melt
water becomes diluted until its freezing point rises to near the freezing point of water.
Freezing then occurs again. Thus, freezing and thawing occur as often as without the
use of de-icing agents, or even more often because a potentially insulating layer of ice
has been destroyed. In consequence, de-icing agents can be said to increase saturation,
and possibly also to increase the number of cycles of freezing and thawing. An
indirect confirmation of this behaviour is offered by the fact that the greatest damage
occurs when concrete is exposed to relatively low concentrations of salts (2 to 4 per
cent solution)11.13 (Fig. 11.15).

Fig. 11.15. Effect of concentration of CaCI2 on scaling of non-air-entrained

concrete after 50 cycles of freezing and thawing (without removal of the
solution).11.13 The extent of surface scaling is rated from 0 = no scaling to 5 =
severe scaling
An additional factor contributing to the damage of concrete is the sudden drop in
temperature of the subsurface concrete when ice melts and extracts the latent heat;
this is a form of thermal shock which can result in very rapid freezing.
Air entrainment makes concrete very much more resistant to surface scaling in the
same way as it provides resistance to freezing and thawing without the use of de-icing
agents. The concrete should have a water/cement ratio not higher than 0.40 and a

610
cement content of at least 310 kg/m3 (520 lb/yd3).11.56 High-strength concrete shows
very good resistance to scaling.11.61
Numerous tests on salt scaling have shown that the extent of damage is sensitive to
the procedure adopted. For instance, air drying of the concrete after wet curing but
prior to exposure cycles, increases the resistance to surface scaling.11.31 The drying out
must, however, be preceded by moist curing of sufficient duration for the cement
paste to hydrate extensively. Therefore, in practice, concreting should take place at a
time of the year such that good curing can be applied, followed by a period of drying
out. Excessive bleeding and laitance must be avoided.
The most severe damage occurs when concrete is subjected to alternating freezing
and thawing with the de-icer solution remaining on top of the specimen, rather than
being replaced with fresh water prior to each re-freezing.11.13 On the other hand, if the
liquid is removed from the surface of the concrete prior to re-freezing, no scaling
takes place, even with non-air-entrained concrete.11.13
The resistance of concrete to de-icing agents can be ascertained by the test method
of ASTM C 672-03 in which specimens are subjected to cycles of freezing when
covered with a calcium chloride solution followed by thawing in air. The assessment
of scaling is made visually.
Because chlorides which penetrate to the reinforcing steel lead to corrosion, the
use of chloride-free de-icing agents is desirable. One of these is urea, which, however,
pollutes water and is less effective in removing ice. Calcium magnesium acetate is
effective, albeit slow to act, but is very expensive.
Some protection of concrete from the deleterious action of de-icing agents can be
obtained by sealing it with linseed oil. Boiled linseed oil, diluted in equal parts with
kerosene or mineral spirits, is applied to the surface of concrete, which must be dry,
two coats being used. The oil slows down the ingress of the de-icer solution but does
not seal the surface of the concrete so as to prevent evaporation. The linseed oil
darkens the colour of concrete, and non-uniform application may produce an
unsightly surface. Re-sealing after a few years is necessary. Silane and siloxane can
also be used, but this is a specialized topic.

Chloride attack*
*
The sections on chloride attack of reinforced concrete were substantially
published in ref. 11.37.
Chloride attack is distinct in that the primary action is the corrosion of steel
reinforcement, and it is only as a consequence of this corrosion that the surrounding
concrete is damaged. Corrosion of reinforcement is one of the major causes of
deterioration of reinforced concrete structures in many locations. The broad topic of
corrosion of steel, as well as of other metals, embedded in concrete (see ACI 222R-
89)11.82 is outside the scope of the present book, and it is intended to limit this
discussion to the consideration of those properties of concrete which influence
corrosion, with emphasis on the transport of chloride ions through the concrete in the
cover to the reinforcement.
Nevertheless, a brief description of the mechanism of chloride-induced corrosion
will be helpful in understanding the processes involved.
Mechanism of chloride-induced corrosion

611
The protective passivity layer on the surface of embedded steel was mentioned on
p. 499. This layer, which is self-generated soon after the hydration of cement has
started, consists of γ-Fe2O3 tightly adhering to the steel. As long as that oxide film is
present, the steel remains intact. However, chloride ions destroy the film and, in the
presence of water and oxygen, corrosion occurs. Chloride ions were described by
Verbeck11.63 as “a specific and unique destroyer”.
It may be useful to add that, provided the surface of the reinforcing steel is free
from loose rust (a condition which is always specified), the presence of rust at the
time when the steel is embedded in concrete does not influence corrosion.11.78
A brief description of the corrosion phenomenon is as follows. When there exists a
difference in electrical potential along the steel in concrete, an electrochemical cell is
set up: there form anodic and cathodic regions, connected by the electrolyte in the
form of the pore water in the hardened cement paste. The positively charged ferrous
ions Fe++ at the anode pass into solution while the negatively charged free electrons e–
pass through the steel into the cathode where they are absorbed by the constituents of
the electrolyte and combine with water and oxygen to form hydroxyl ions (OH)–.
These travel through the electrolyte and combine with the ferrous ions to form ferric
hydroxide which is converted by further oxidation to rust (see Fig. 11.16). The
reactions involved are as follows:
anodic reactions:

cathodic reaction:

4e- + O2 + 2H2O → 4(OH)-.

612
 Fig. 11.16. Schematic representation of electro-chemical corrosion in the
presence of chlorides
It can be seen that oxygen is consumed and water is regenerated but it is needed
for the process to continue. Thus, there is no corrosion in, dry concrete, probably
below a relative humidity of 60 per cent; nor is there corrosion in concrete fully
immersed in water, except when water can entrain air, for example by wave action.
The optimum relative humidity for corrosion is 70 to 80 per cent. At higher relative
humidities, the diffusion of oxygen through the concrete is considerably reduced.
The differences in electrochemical potential can arise from differences in the
environment of the concrete, for example when a part of it is permanently submerged
in sea water and a part is exposed to periodic wetting and drying. A similar situation
can arise when there is a substantial difference in the thickness of cover to a steel
system which is electrically connected. Electrochemical cells form also due to a
variation in salt concentration in the pore water or due to a non-uniform access to
oxygen.
For corrosion to be initiated, the passivity layer must be penetrated. Chloride ions
activate the surface of the steel to form an anode, the passivated surface being the
cathode. The reactions involved are as follows:

Fe++ + 2Cl– → FeCl2

FeCl2 + 2H2O → Fe(OH)2 + 2HCl.

Thus, Cl– is regenerated so that the rust contains no chloride, although ferrous chloride
is formed at the intermediate stage.
Because the electrochemical cell requires a connection between the anode and the
cathode by the pore water, as well as by the reinforcing steel itself, the pore system in
hardened cement paste is a major factor influencing corrosion. In electrical terms, it is

613
the resistance of the ‘connection’ through the concrete that controls the flow of the
current. The electrical resistivity of concrete is greatly influenced by its moisture
content, by the ionic composition of the pore water, and by the continuity of the pore
system in the hardened cement paste.
There are two consequences of corrosion of steel. First, the products of corrosion
occupy a volume several times larger than the original steel so that their formation
results in cracking (characteristically parallel to the reinforcement), spalling or in
delamination of concrete (see Fig. 11.17). This makes it easier for aggressive agents
to ingress toward the steel, with a consequent increase in the rate of corrosion. Second,
the progress of corrosion at the anode reduces the cross-sectional area of the steel,
thus reducing its load-carrying capacity. In this connection, it should be pointed out
that chloride-induced corrosion is highly localized at a small anode, pitting of the steel
taking place.

Fig. 11.17. Diagrammatic representation of damage induced by corrosion:

cracking, spalling, and delamination

614
 When the supply of oxygen is severely limited, corrosion at a slow rate can occur.
The products of corrosion, which are less voluminous than under normal
circumstances, may travel into voids in the concrete without a progressive
development of cracking or spalling.
Chlorides in the mix
Chlorides can be present in concrete because they have been incorporated in the mix
through the use of contaminated aggregate or of sea water or brackish water, or by
admixtures containing chlorides. None of these materials should be permitted in
reinforced concrete, and standards generally prescribe strict limits on the chloride
content of the concrete from all sources. For example, BS 8110-1 : 1997 limits
the total chloride-ion content in reinforced concrete to 0.40 per cent by mass of
cement. The same limit is prescribed by European Standard BS EN 206-1 : 2000 (now
BS EN 1992-1 : 2004). The approach of ACI 318-0211.56 is to consider water-
soluble chloride ions only. On that basis, the chloride-ion content of reinforced
concrete is limited to 0.15 per cent by mass of cement. The two values are not
substantially different from one another because water-soluble chlorides are only a
part of the total chloride content, namely, the free chlorides in pore water. The
distinction between free and bound chlorides is considered on p. 571, but, at this stage,
it can be noted that the total chloride content is determined as the acid-
soluble chloride content, using ASTM C 1152-04 or BS 1881-124 : 1988. In the
presence of some admixtures, potentiometric titration gives a higher value of chloride
content than reliance on colour change. There exist several techniques for the
determination of the content of water-soluble chlorides.
As a possible source of chlorides in the mix, Portland cement itself contains only a
very small amount: normally, no more than 0.01 per cent by mass. However, ground
granulated blastfurnace slag may have a significant chloride content if its processing
involved quenching with sea water.11.92 Drinking water may well contain 250 ppm of
chloride ions; at a water/cement ratio of 0.4, the water would contribute the same
amount of chloride ions as Portland cement. As far as aggregate is concerned, BS 882 :
1992 (withdrawn) gives guidance on the maximum total chloride ion content;
compliance with this guidance is likely to satisfy the requirements for concrete of BS
5328-1 : 1997 (withdrawn) and of BS 8110-1 : 1997 (now Eurocode 2 : 2004). For
reinforced concrete, the chloride content of the aggregate should not exceed 0.05 per
cent by mass of the total aggregate; this is reduced to 0.03 per cent when sulfate-
resisting cement is used. For prestressed concrete, the corresponding figure is 0.01 per
cent. Limits on impurities in water are given in BS EN 1008 : 2002 and ASTM C
1602-06.
The various limits on chlorides referred to in this section are generally
conservative so that compliance with them should ensure no chloride-induced
corrosion unless more chlorides ingress into the concrete in service. The view that the
limits are conservative is disputed by Pfeifer.11.40
Ingress of chlorides
The problem of chloride attack arises usually when chloride ions ingress from outside.
This can be caused by de-icing salts – a topic discussed on p. 563. Another,
particularly important, source of chloride ions is sea water in contact with concrete.
Chlorides can also be deposited on the surface of concrete in the form of air-borne
very fine droplets of sea water (raised from the sea by turbulence and carried by wind)

615
or of air-borne dust which subsequently becomes wetted by dew. It is useful to point
out that air-borne chlorides can travel substantial distances: 2 km has been
reported,11.75 but travel over even greater distances is possible, depending on wind and
topography. The configuration of structures also affects the movement of air-borne
salts: when eddies occur in the air, salts can reach the landward faces of structures.
Brackish groundwater in contact with concrete is also a source of chlorides.
Although this is a rare occurrence, it may be mentioned that chlorides can ingress
into concrete from conflagration of organic materials containing chlorides.
Hydrochloric acid is formed and deposited on the surface of concrete where it reacts
with calcium ions in the pore water. Ingress of chloride ions can follow.11.83
Whatever their external origin, chlorides penetrate concrete by transport of water
containing the chlorides, as well as by diffusion of the ions in the water, and by
absorption. Prolonged or repeated ingress can, with time, result in a high
concentration of chloride ions at the surface of the reinforcing steel.
When concrete is permanently submerged, chlorides ingress to a considerable
depth but, unless oxygen is present at the cathode, there will be no corrosion. In
concrete which is sometimes exposed to sea water and is sometimes dry, the ingress
of chlorides is progressive. The following is a description of a situation often found in
structures on the coast in a hot climate.
Dry concrete imbibes salt water by absorption and, under some conditions, may
continue to do so until the concrete has become saturated. If the external conditions
then change to dry, the direction of movement of water becomes reversed and water
evaporates from the ends of capillary pores open to the ambient air. It is, however,
only pure water that evaporates, the salts being left behind. Thus, the concentration of
salts in the water still in the concrete increases near the surface of the concrete. The
concentration gradient thus established drives the salts in the water near the surface of
the concrete towards the zones of lower concentration, i.e. inwards; this is transport
by diffusion. Depending on the external relative humidity and on the duration of the
drying period, it is possible for most of the water in the outer zone of the concrete to
evaporate so that the water remaining in the interior will become saturated with salt
and the excess salt will precipitate out as crystals.
It can be seen thus that, in effect, the water moves outwards and the salt inwards.
The next cycle of wetting with salt water will bring more salt present in solution into
the capillary pores. The concentration gradient now decreases outwards from a peak
value at a certain depth from the surface, and some salts may diffuse toward the
surface of the concrete. If, however, the wetting period is short and drying restarts
quickly, the ingress of salt water will carry the salts well into the interior of the
concrete; subsequent drying will remove pure water, leaving the salts behind.
The exact extent of the movement of salt depends on the length of the wetting and
drying periods. It may be recalled that wetting of concrete occurs very rapidly and
drying is very much slower; the interior of the concrete never dries out. It should also
be noted that the diffusion of ions during the wet periods is fairly slow.
It is apparent thus that a progressive ingress of salts toward the reinforcing steel
takes place under alternating wetting and drying, and a chloride profile of the kind
shown in Fig. 11.18 is established. The profile is determined by chemical analysis of
dust samples obtained by incremental drilling to various depths from the surface.
Sometimes, there is a lower concentration of chlorides in the outermost 5 mm ( in.)

616
or so of the concrete where rapid movement of water takes place so that the salts are
quickly carried a small distance inwards. The maximum chloride ion content in pore
water can be in excess of the concentration in sea water; this was observed after 10
years’ exposure.11.71 The crucial fact is that, with the passage of time, a sufficient
amount of chloride ions will reach the surface of the reinforcing steel. What
constitutes a ‘sufficient’ amount will be discussed in the subequent section.

Fig. 11.18. An example of the profile of total chloride ion content as a percentage
of the mass of cement; points show averages over 10 or 20 mm increments
As just mentioned, the ingress of chlorides into concrete is strongly influenced by
the exact sequence of wetting and drying. This sequence varies from location to
location, depending on the movement of the sea and on the wind, on exposure to the
sun, and on the usage of the structure. Thus, even different parts of the same structure
may undergo a different pattern of wetting and drying; this explains why, sometimes,
there is a considerable variation in the extent of corrosion damage in a single structure.
It is not only wetting and drying of the surface zone of the concrete that influences
the ingress of chlorides; drying to a greater depth allows subsequent wetting to carry
the chlorides well into the concrete, thus speeding up the ingress of chloride ions. For
this reason, concrete in the tidal zone (where the period of drying is short) is less
vulnerable to corrosion than concrete in the splash zone (where wetting may occur
only when the sea is high or the wind is strong). The most vulnerable is the concrete
wetted by sea water only occasionally, such as areas around bollards (where wet ropes
are coiled) or in the vicinity of fire hydrants (using sea water), or in industrial areas
subjected to periodic washdown with sea water, but at other times exposed to the
drying effects of the sun and of a high temperature.
Threshold content of chloride ions

617
It was mentioned earlier that, for corrosion to be initiated, there has to be a certain
minimum concentration of chloride ions at the surface of the steel. However, no
universally valid threshold concentration exists. As far as chlorides incorporated in
the original mix are concerned, the threshold concentration was considered on p. 568.
It is useful to add that the presence of a given excessive amount of chlorides in the
original mix results in a more aggressive action, and therefore a higher corrosion rate,
than when the same amount of chlorides has ingressed into the concrete in service.11.64
As far as chlorides which have ingressed into the concrete are concerned, it is even
more difficult to establish a threshold concentration of chloride ions below which
there is no corrosion. This threshold depends on a number of factors, many of which
are still imperfectly understood. Moreover, the distribution of chlorides within the
hardened cement paste is not uniform, as found in chloride profiles in actual structures.
For practical purposes, prevention of corrosion lies in controlling the ingress of
chlorides by the thickness of cover to reinforcement and by the penetrability of the
concrete in the cover.
While, under any given circumstances, there may be a threshold chloride content
for corrosion to be initiated, its progress depends on the resistivity of the hardened
cement paste, which varies with humidity, and on the availability of oxygen, which is
affected by the immersion of concrete.
In any case, it is not the total chloride content that is relevant to corrosion. A part
of the chlorides are chemically bound, being incorporated in the products of hydration
of cement. Another part of the chlorides are physically bound, being adsorbed on the
surface of the gel pores. It is only the third part of the chlorides, namely, free
chlorides, that are available for the aggressive reaction with steel. However, the
distribution of the chloride ions among the three forms is not permanent as there is an
equilibrium situation such that some free chloride ions are always present in the pore
water. It follows that only the chloride ions in excess of those needed for this
equilibrium can become bound.
Binding of chloride ions
The main form of binding of the chloride ions is by reaction with C3A to form calcium
chloroaluminate, 3CaO.Al2O3.CaCl2.10H2O, sometimes referred to as Friedel’s salt. A
similar reaction with C4AF results in calcium chloroferrite, 3CaO.Fe2O3.CaCl2.10H2O.
It follows that more chloride ions are bound when the C3A content of the cement is
higher, and also when the cement content of the mix is higher. For this reason, it used
to be thought that cements with a high C3A content are conducive to good resistance
to corrosion.
This may be true when chloride ions are present at the time of mixing (a situation
which should not be permitted) because they can rapidly react with C3A. However,
when chloride ions ingress into concrete, a smaller amount of chloroaluminates is
formed and, under some future circumstances, they may become dissociated,
releasing chloride ions so as to replenish those removed from the pore water by
transport to the surface of the steel.
A further factor in deciding on the desirable C3A content of the cement is the
possibility of sulfate attack on some parts of the given structure, other than
those subject to the ingress of sea water. As mentioned on p. 76, sulfate resistance
requires a low C3A content in the cement. For these various reasons, it is nowadays
thought that a moderately sulfate-resisting cement, Type II, offers the best
compromise.

618
 In the case of cements containing ground granulated blastfurnace slag, it has been
suggested that binding of chlorides takes place also by the aluminates in the slag, but
this has not been fully confirmed.11.91
In connection with a possible use of cement with a high C3A content, it should be
remembered that a high C3A content results in a higher early rate of heat evolution,
and therefore a temperature rise. This behaviour can be harmful in moderately large
concrete masses often associated with structures exposed to the sea.11.88
Some standards, for example BS 8110-1 : 1985 (replaced by Eurocode 2-2004),
severely limit the chloride content when sulfate-resisting cement (Type V) is used, on
the assumption that chlorides adversely affect sulfate resistance. This has now been
proven not to be the case.11.76 What happens is that sulfate attack results in a
decomposition of calcium chloroaluminate, thus making some chloride ions available
for corrosion; calcium sulfoaluminate is formed.11.79
Carbonation of hardened cement paste in which bound chlorides are present has a
similar effect of freeing the bound chlorides and thus increasing the risk of corrosion.
Ho and Lewis11.80 cite Tuutti as having found an increased concentration of chloride
ions in pore water to occur 15 mm in advance of the carbonation front. This harmful
effect of carbonation is in addition to the lowering of the pH value of the pore water,
so that severe corrosion may well follow. It has also been found in laboratory
tests11.85 that the presence of even a small amount of chlorides in carbonated concrete
enhances the rate of corrosion induced by the low alkalinity of carbonated concrete.
In considering both carbonation and ingress of chloride ions, it is important to
remember that the optimum relative humidity for carbonation is between 50 and 70
per cent, whereas corrosion progresses rapidly only at higher humidities. The
occurrence of both of these relative humidities, one after another, is possible when
concrete is exposed to long periods of alternating wetting and drying. Another
occurrence of both chloride ingress and carbonation was observed in thin cladding
panels of a building: air-borne chlorides ingressed from outside and reached the
reinforcing steel; carbonation progressed from the relatively dry inside of the building.
Returning to the topic of the chloride ion concentration present in the pore water in
an equilibrium situation, it should be noted that the chloride ion concentration
depends on the other ions present in the pore water; for example, at a given total
chloride ion content, the higher the hydroxyl (OH–) concentration the more free
chloride ions are present.11.66 For this reason, the Cl–/OH– ratio is considered to affect
the progress of corrosion, but no generally valid statements can be made. It has also
been found that, for given amount of chloride ions in the mix, there are significantly
more free chloride ions with NaCl than with CaCl2.11.67
Because of these various factors, the proportion of bound chloride ions varies from
80 per cent to well below 50 per cent of the total chloride ion content. Therefore, there
may not exist a fixed and unique value of the total amount of chloride ions below
which corrosion will not occur. Tests11.66,11.68 have shown that, in consequence of the
various equilibrium requirements of the pore water, the mass of bound chlorides in
relation to the mass of cement is independent of the water/cement ratio.
Influence of blended cements on corrosion
While the preceding discussion was concerned with the influence of the type of
Portland cement on the chemical aspects of chloride ions, it is also important, indeed
more so, to consider the influence of the type of blended cement on the pore structure

619
of the hardened cement paste and on its penetrability, as well as on resistivity. This
was largely done in Chapter 10, but those aspects of various cementitious materials
which are particularly relevant to the movement of chloride ions will be considered
here. It should be added that the same properties of hardened cement paste which
influence the transport of chlorides also influence the supply of oxygen and the
availability of moisture, both of which are necessary for corrosion to occur. However,
the locations on steel where chlorides are present and where oxygen is needed are
different: the former is at the anode, and the latter at the cathode.
The cementitious materials of interest are fly ash, ground granulated blast-furnace
slag, and silica fume. All three, when properly proportioned in the mix, significantly
reduce the penetrability of concrete and increase its resistivity, thereby reducing the
rate of corrosion.11.70,11.87,11.90 As far as silica fume is concerned, its positive effect is
through improvement of the pore structure of hardened cement paste, which increases
resistivity, even though silica fume reduces somewhat the pH value of the pore water
in consequence of reaction with Ca(OH)2.11.98 Gjørv et al.11.97 showed that 9 per cent of
silica fume by mass in the cement reduced the chloride diffusivity by a factor of about
5.
It should be remembered that, because of its effect on workability, with the use of
silica fume there is usually associated the inclusion of a superplasticizer.
Superplasticizers per se do not affect the pore structure and, therefore, do not alter the
process of corrosion.
The beneficial effects of the various cementitious materials are so significant that
their use in reinforced concrete liable to corrosion in hot climates is virtually
necessary: Portland cement alone should not be used.11.89
Tests on chloride ion diffusion through mortar indicate that fillers do not affect the
movement of chlorides.11.77
Chloride ions in concrete made with high-alumina cement lead to a more
aggressive situation than with Portland cement,11.81 the comparison being made at the
same chloride ion content. It can be recalled that the pH value in high-alumina cement
concrete is lower than with Portland cement so that the passive state of the steel may
be less stable.11.81
Further factors influencing corrosion
The preceding discussion of the influence of the composition of concrete upon its
resistance to corrosion should be complemented by re-emphasizing the importance of
good curing, whose effect is primarily upon the concrete in the cover zone. The time
to initiation of corrosion is substantially increased by prolonged curing11.69 (see Fig.
11.19). However, only fresh water must be used for curing because brackish water
greatly increases the ingress of chlorides.11.69

620
 Fig. 11.19. Influence of the length of moist curing on time to initiation of
corrosion of reinforcement; water/cement ratio of 0.5, cement content 330
kg/m3 (550 lb/yd2), Type V cement; specimens partially immersed in a 5 per cent
solution of sodium chloride (based on ref. 11.69)
Once the corrosion has been initiated, its continuation is not inevitable: the
progress of corrosion is influenced by the resistivity of the concrete between the
anode and the cathode and by the continuing supply of the oxygen at the cathode. On
the one hand, it is very doubtful that the supply of oxygen can be completely and
reliably stopped by the application of a membrane, although developments in this
field continue. On the other hand, the resistivity of concrete is a function of its
moisture condition so that drying out would halt the corrosion, which, however, can
re-start upon subsequent wetting.
Cracking of concrete in the cover facilitates the ingress of chlorides and, therefore,
enhances corrosion. Although virtually all reinforced concrete in service exhibits
some cracks, cracking can be controlled by appropriate structural design, detailing,
and construction procedures. Cracks wider than about 0.2 to 0.4 mm (0.008 to 0.016
in.) are harmful. It may be worth mentioning that, although prestressed concrete is
crack-free, the prestressing steel is more vulnerable to corrosion because of its nature;
also, the small cross-sectional area of prestressing wires means that pitting corrosion
greatly reduced their load-carrying capacity.
Higher temperature has several effects on corrosion. First, the content of free
chlorides in the pore water increases; the effect is more pronounced with

621
cements having a high C3A content and with lower chloride concentrations in the
original mix11.62 (Fig. 11.20).

Fig. 11.20. Influence of C3A content in cement on the amount of free chloride ions
(expressed as a percentage of total chloride ions of 1.2 per cent of the mass of
cement) at 20 and 70 °C (68 and 158 °F) (based on ref. 11.62 with kind
permission of Elsevier Science Ltd, Kidlington, U.K.)
More importantly, the reactions of corrosion, like many chemical reactions, occur
faster at higher temperatures. It is usually assumed that a rise of temperature of 10 °C
(18 °F) doubles the rate of reaction, but there is some evidence that the increase is
only 1.6-fold.11.93 Whatever the exact factor, the accelerating effect of temperature
explains why there is so much more corrosion-damaged concrete in hot coastal areas
than in temperate parts of the world.
It may also be recalled that initial hardening of concrete at high temperatures
results in a coarser pore structure (see p. 361), a consequence of which is a lower
resistance to the diffusion of chloride ions.11.39 The temperature differential between the
surface of the concrete and its interior affects the diffusion; direct exposure to the sun
can result in a significant rise in the temperature of the surface concrete above the
ambient value.
Thickness of cover to reinforcement
The thickness of cover to reinforcement is an important factor controlling the
transport of chloride ions: the greater the cover the longer the time interval before the
chloride ion concentration at the surface of the steel reaches the threshold value. Thus,
the quality of the concrete (in terms of its low penetrability) and the thickness of
cover work together and can, therefore, to some extent, be traded off one against the
other. For this reason, standards often specify combinations of cover and strength of
concrete such that a lower thickness of cover requires a higher strength, and vice
versa.
However, there are limitations to this approach. First of all, thick cover is of no
avail if the concrete is highly penetrable. Moreover, the purpose of cover is not only

622
to provide protection of reinforcement, but also to ensure composite structural action
of steel and concrete, as well as, in some cases, to provide fire protection or resistance
to abrasion. Unduly large thickness of cover would result in the presence of a
considerable volume of concrete devoid of reinforcement. And yet, the presence of
steel is required to control shrinkage and thermal stresses, and to prevent cracking due
to those stresses. Were cracking to occur, the large thickness of cover would be
proved to be detrimental. In practical terms, the cover thickness should not exceed 80
to 100 mm (3 to 4 in.) but the decision on cover forms part of structural design.
Too small a thickness of cover should not be used either, because, however low the
penetrability of the concrete, cracking, for whatever reason, or local damage or
misplaced reinforcement can result in a situation where chloride ions can rapidly be
transported to the surface of the steel.
Tests for penetrability of concrete to chlorides
A rapid test for the penetrability of concrete to chloride ions is prescribed by ASTM C
1202-10, which determines the electrical conductance, expressed as the total electrical
charge in coulombs (ampere-seconds) passed during a certain time interval through a
concrete disc between solutions of sodium chloride and sodium hydroxide when a
potential difference of 60 V d.c. is maintained. The charge is related to the
penetrability of the concrete to chloride ions, so that the test can be of help, in a
comparative manner, in selecting a suitable concrete mix. A somewhat similar test
determines the a.c. impedance of specimens of various shapes.11.86
Tests of the kind just described do not necessarily replicate the transport of
chloride ions in a real-life situation, nor do they have a sound scientific basis.
Nevertheless, they are useful and certainly preferable to the assumption that resistance
to chloride ion ingress is simply related to the strength of concrete; this assumption
has been shown not to be valid11.41 except in the most general manner.
Stopping corrosion
Simplified statements about methods of controlling or remedying corrosion which has
been initiated may be unhelpful. All that should be stated here is that the progress of
corrosion would be reduced by drying the concrete or by the prevention of oxygen
supply through the application of surface barriers. This is a specialized field, and ad
hoc solutions may, in fact, prove harmful; for instance, applying a barrier at the anode
(rather than the cathode) would increase the ratio of the size of the cathode to the
anode, which would increase the rate of corrosion.
It is reasonable to raise the question of whether there exist integral corrosion
inhibitors, that is, substances which, while not preventing ingress of chlorides into
concrete, inhibit the corrosion of steel. Nitrites of sodium11.74 and calcium11.72 have been
found to be effective in laboratory tests. The action of the nitrite is to convert ferrous
ions at the anode into a stable passive layer of Fe2O3, the nitrite ion reacting
preferentially to the chloride ion. The concentration of nitrites must be sufficient to
cope with a continuing ingress of chloride ions. Indeed, it is not certain that corrosion
inhibitors are effective indefinitely, and do not simply delay corrosion.
The accelerating effect of the nitrites can be offset by the use of a retarding
admixture, if need be. The search for other corrosion inhibitors continues.11.73
Being incorporated in the mix, the inhibitors protect all the embedded steel.
Nevertheless, inhibitors are no substitute for concrete of low penetrability: they are

623
merely an additional safeguard. Moreover, sodium nitrite increases the hydroxyl ion
concentration in the pore water, and this may increase the risk of alkali–aggregate
reaction. Thus, the beneficial effect of an increased hydroxyl ion concentration upon
the risk of corrosion of steel is accompanied by a negative effect on the risk of alkali–
aggregate reaction. Of course, this is relevant only if the aggregate is susceptible to
such a reaction in the first place.
A discussion of prevention of the corrosion of steel in concrete would be
incomplete without a mention of the protection of steel by epoxy coating and by
cathodic protection which makes the entire steel surface cathodic. Epoxy coating of
steel is a specialized technique which can be helpful in addition to an adequate
thickness of cover concrete of low permeability. In special cases, reinforcement made
of stainless steel, or coated with stainless steel, can be used, but this is very expensive.
Cathodic protection has been shown to be effective in some applications, but its use in
a new structure is an admission of defeat in that the particular reinforced concrete
structure is manifestly not durable.
A question which has to be faced occasionally is: can chloride ions be removed
from the surface of the steel. Within the confines of this book, only a very brief
answer can be given.
There has been developed a technique for desalinating concrete, in which chloride
is removed by passing a heavy direct current between the corroding reinforcing steel
(now acting as a cathode) and an external anode in electrolytic contact with the
concrete; chloride ions migrate towards the external anode, thus moving away from
the surface of the reinforcement.11.84 It seems that only about one-half of the chloride in
the concrete can be removed and, with time, corrosion is likely to re-start. Some
negative consequences of the process may follow;11.65 for example, the concentration of
sodium ions which enter the pore water can become so high that aggregate which,
under normal circumstances, is non-reactive with the alkalis may become reactive.
References
11.1. T. C. POWERS, L. E. COPELAND and H. M. MANN, Capillary continuity or
discontinuity in cement pastes, J. Portl. Cem. Assoc, Research and
Development Laboratories, 1, No. 2, pp. 38–48 (May 1959).
11.2. CENTRE D’INFORMATION DE L’INDUSTRIE CIMENTIÈRE BELGE, Le béton et le gel, Bull.
No. 61 to 64 (Sept. to Dec. 1957).
11.3. G. MOLLER, Tests of resistance of concrete to early frost action, RILEM
Symposium on Winter Concreting (Copenhagen, 1956).
11.4. T. C. POWERS, Resistance to weathering – freezing and thawing, ASTM Sp.
Tech. Publ. No. 169, pp. 182–7 (1956).
11.5. T. C. POWERS, What resulted from basic research studies, Influence of Cement
Characteristics on the Frost Resistance of Concrete, pp. 28–43 (Chicago,
Portland Cement Assoc., Nov. 1951).
11.6. R. A. HELMUTH, Capillary size restrictions on ice formation in hardened
portland cement pastes, Proc. 4th Int. Symp. on the Chemistry of Cement,
Washington DC, pp. 855–69 (1960).
11.7. A. R. COLLINS, Discussion on: A working hypothesis for further studies of
frost resistance of concrete by T. C. Powers, J. Amer. Concr. Inst., 41,
(Supplement) pp. 272-12–14 (Nov. 1945).

624
 11.8. T. C. POWERS, Some observations on using theoretical research. J. Amer.
Concr. Inst., 43, pp. 1089–94 (June 1947).
11.9. G. J. VERBECK, What was learned in the laboratory, Influence of Cement
Characteristics on the Frost Resistance of Concrete, pp. 14–27 (Chicago,
Portland Cement Assoc., Nov. 1951).
11.10. U.S. BUREAU OF RECLAMATION, Relationship of moisture content of aggregate to
durability of the concrete, Materials Laboratories Report No. C-513 (Denver,
Colorado, 1950).
11.11. U.S. BUREAU OF RECLAMATION, Investigation into the effect of water/cement
ratio on the freezing–thawing resistance of non-air and air-entrained
concrete, Concrete Laboratory Report No. C-810 (Denver, Colorado, 1955).
11.12. G. J. VERBECK and P. KLIEGER, Calorimeter-strain apparatus for study of
freezing and thawing concrete, Highw. Res. Bd Bull. No. 176, pp. 9–12
(Washington DC, 1958).
11.13. G. J. VERBECK and P. KLIEGER, Studies of “salt” scaling of concrete, Highw.
Res. Bd Bull. No. 150, pp. 1–13 (Washington DC, 1957).
11.14. P. KLIEGER, Further studies on the effect of entrained air on strength and
durability of concrete with various sizes of aggregates, Highw. Res. Bd Bull.
No. 128, pp. 1–19 (Washington DC, 1956).
11.15. T. C. POWERS, Void spacing as a basis for producing air-entrained concrete, J.
Amer. Concr. Inst., 50, pp. 741–60 (May 1954), and Discussion, pp. 760-6–
15 (Dec. 1954).
11.16. U.S. BUREAU OF RECLAMATION, The air-void systems of Highway Research
Board co-operative concretes, Concrete Laboratory Report No. C-
824 (Denver, Colorado, April 1956).
11.17. T. C. POWERS and R. A. HELMUTH, Theory of volume changes in hardened
portland cement paste during freezing, Proc. Highw. Res. Bd, 32, pp. 285–97
(Washington DC, 1953).
11.18. P. J. F. WRIGHT, Entrained air in concrete, Proc. Inst. Civ. Engrs., Part 1, 2,
No. 3, pp. 337–58 (London, May 1953).
11.19. L. S. BROWN and C. U. PIERSON, Linear traverse technique for measurement of
air in hardened concrete, J. Amer. Concr. Inst., 47, pp. 117–23 (Oct. 1950).
11.20. T. C. POWERS, Basic considerations pertaining to freezing and thawing
tests, Proc. ASTM, 55, pp. 1132–54 (1955).
11.21. HIGHWAY RESEARCH BOARD, Report on co-operative freezing and thawing tests
of concrete, Special Report No. 47 (Washington DC, 1959).
11.22. H. WOODS, Observations on the resistance of concrete to freezing and
thawing, J. Amer. Concr. Inst., 51, pp. 345–9 (Dec. 1954).
11.23. J. VUORINEN, On the use of dilation factor and degree of saturation in testing
concrete for frost resistance, Nordisk Betong, No. 1, pp. 37–64 (1970).
11.24. M. A. WARD, A. M. NEVILLE and S. P. SINGH, Creep of air-entrained
concrete, Mag. Concr. Res., 21, No. 69, pp. 205–10 (Dec. 1969).
11.25. D. STARK, Characteristics and utilization of coarse aggregates associated with
D-cracking, ASTM Sp. Tech. Publ. No. 597, pp. 45–58 (1976).

625
 11.26. C. MACINNIS and J. D. WHITING, The frost resistance of concrete subjected to a
deicing agent, Cement and Concrete Research, 9, No. 3, pp. 325–35 (1979).
11.27. B. MATHER, Tests of high-range water-reducing admixtures,
in Superplasticizers in Concrete, ACI SP-62 pp. 157–66 (Detroit, Michigan,
1979).
11.28. R. D. GAYNOR and J. I. MULLARKY, Effects of mixing speed on air
content, NRMCA Technical Information Letter No. 312 (Silver Spring,
Maryland, National Ready Mixed Concrete Assoc., Sept. 20, 1974).
11.29. H. SOMMER, Ein neues Verfahren zur Erzielung der Frost-Tausalz-
Beständigkeit des Betons, Zement und Beton, 22, No. 4, pp. 124–9 (1977).
11.30. B. MATHER, Concrete need not deteriorate, Concrete International, 1, No. 9,
pp. 32–7 (1979).
11.31. B. MATHER, A discussion of the paper “Mechanism of the CaCl2 attack on
Portland cement concrete”, by S. CHATTERJI, Cement and Concrete
Research, 9, No. 1, pp. 135–6 (1979).
11.32. L. H. TUTHILL, Resistance to chemical attack, ASTM Sp. Tech. Publ. No. 169B,
pp. 369–87 (1978).
11.33. R. D. GAYNOR, Ready-mixed concrete, ASTM Sp. Tech. Publ. No. 169B, pp.
471–502 (1978).
11.34. M. PIGEON, La durabilité au gel du béton, Materials and Structures, 22, No.
127, pp. 3–14 (1989).
11.35. M. PIGEON, P.-C. AÏTCIN and P. LAPLANTE, Comparative study of the air-void
stability in a normal and a condensed silica fume field concrete, ACI
Materials Journal, 84, No. 3, pp. 194–9 (1987).
11.36. R. C. PHILLEO, Freezing and Thawing Resistance of High Strength Concrete,
Report 129, Transportation Research Board, National Research Council, 31
pp. (Washington DC, 1986).
11.37. A. NEVILLE, Chloride attack of reinforced concrete – an overview, Materials
and Structures, 28, No. 176, pp. 63–70 (1995).
11.38. J. T. HOARTY, Improved air-entraining agents for use in concretes containing
pulverised fuel ashes, in Admixtures for Concrete: Improvement of
Properties, Proc. ASTM Int. Symposium, Barcelona, Spain, Ed. E. Vázquez,
pp. 449–59 (London, Chapman and Hall, 1990).
11.39. R. J. DETWILER, K. O. KJELLSEN and O. E. GJØRV, Resistance to chloride
intrusion of concrete cured at different temperatures, ACI Materials
Journal, 88, No. 1, pp. 19–24 (1991).
11.40. D. W. PFEIFER, W. F. PERENCHIO and W. G. HIME, A critique of the ACI 318
chloride limits, PCI Journal, 37, No. 5, pp. 68–71 (1992).
11.41. H. R. SAMAHA and K. C. HOVER, Influence of microcracking on the mass
transport properties of concrete, ACI Materials Journal, 89, No. 4, pp. 416–
24 (1992).
11.42. G. FAGERLUND, Air-pore instability and its effect on the concrete
properties, Nordic Concrete Research, No. 9, pp. 39–52 (Oslo, Dec. 1990).
11.43. M. A. ALI, A Review of Swedish Concreting Practice, Building Research
Establishment Occasional Paper, 35 pp. (Watford, U.K., June 1992).

626
 11.44. F. SAUCIER, M. PIGEON and G. CAMERON, Air-void stability, Part V: temperature,
general analysis and performance index, ACI Materials Journal, 88, No. 1,
pp. 25–36 (1991).
11.45. M. PIGEON and P. PLANTE, Study of cement paste microstructure around air
voids: influence and distribution of soluble alkalis, Cement and Concrete
Research, 20, No. 5, pp. 803–14 (1990).
11.46. K. OKKENHAUG and O. E. GJØRV, Effect of delayed addition of air-entraining
admixtures to concrete, Concrete International, 14, No. 10, pp. 37–41
(1992).
11.47. W. F. PERENCHIO, V. KRESS and D. BREITFELLER, Frost lenses? Sure. But in
concrete?, Concrete International, 12, No. 4, pp. 51–3 (1990).
11.48. D. WHITING, Air contents and air-void characteristics in low-slump dense
concretes, ACI Journal, 82, No. 5, pp. 716–23 (1985).
11.49. G. G. LITVAN, Further study of particulate admixtures for enhanced freeze–
thaw resistance of concrete, ACI Journal, 82, No. 5, pp. 724–30 (1985).
11.50. O. E. GJØRV et al., Frost resistance and air-void characteristics in hardened
concrete, Nordic Concrete Research, No. 7, pp. 89–104 (Oslo, Dec. 1988).
11.51. P.-C. AÏTCIN, C. JOLICOEUR and J. G. MACGREGOR, Superplasticizers: how they
work and why they sometimes don’t, Concrete International, 16, No. 5, pp.
45–52 (1994).
11.52. E.-H. RANISCH and F. S. ROST SY, Salt-scaling resistance of concrete with air-
entrainment and superplasticizing admixtures. in Durability of Concrete:
Aspects of Admixtures and Industrial By-Products, Proc. 2nd International
Seminar, D9 : 1989, pp. 170–8 (Stockholm, Swedish Council for Building
Research, 1989).
11.53. C. OZYILDIRIM and M. M. SPRINKEL, Durability of concrete containing hollow
plastic microspheres, ACI Journal, 79, No. 4, pp. 307–11 (1982).
11.54. J. YINGLING, G. M. MULLINS and R. D. GAYNOR, Loss of air content in pumped
concrete, Concrete International, 14, No. 10, pp. 57–61 (1992).
11.55. D. WHITING, G. W. SEEGEBRECHT and S. TAYABJI, Effect of degree of
consolidation on some important properties of concrete, in Consolidation of
Concrete, ACI SP-96, pp. 125–60 (Detroit, Michigan, 1987).
11.56. ACI 318–95, Building code requirements for structural concrete, ACI
Manual of Concrete Practice, Part 3: Use of Concrete in Building–Design,
Specifications, and Related Topics, 345 pp. (Detroit, Michigan, 1996).
11.57. F. SAUCIER, M. PIGEON and P. PLANTE, Air-void stability, Part III: field tests of
superplasticized concretes, ACI Materials Journal, 87, No. 1, pp. 3–11
(1990).
11.58. G. FAGERLUND, Effect of the freezing rate on the frost resistance of
concrete, Nordic Concrete Research, No. 11, pp. 20–36 (Oslo, Feb. 1992).
11.59. M. PIGEON, J. PRÉVOST and J.-M. SIMARD, Freeze–thaw durability versus
freezing rate, ACI Journal, 82, No. 5, pp. 684–92 (1985).
11.60. C. FOY, M. PIGEON and M. BANTHIA, Freeze–thaw durability and deicer salt
scaling resistance of a 0.25 water–cement ratio concrete, Cement and
Concrete Research, 18, No. 4, pp. 604–14 (1988).

627
 11.61. R. GAGNÉ, M. PIGEON and P.-C. AÏTCIN, Deicer salt scaling resistance of high
strength concretes made with different cements, in Durability of Concrete,
Vol. 1, ACI SP-126, pp. 185–99 (Detroit, Michigan, 1991).
11.62. S. E. HUSSAIN and RASHEEDUZZAFAR, Effect of temperature on pore solution
composition in plain concrete, Cement and Concrete Research, 23, No. 6, pp.
1357–68 (1993).
11.63. G. J. VERBECK, Mechanisms of corrosion in concrete, in Corrosion of Metals
in Concrete, ACI SP-49, pp. 21–38 (Detroit, Michigan, 1975).
11.64. P. LAMBERT, C. L. PAGE and P. R. W. VASSIE, Investigations of reinforcement
corrosion. 2. Electrochemical monitoring of steel in chloride-contaminated
concrete, Materials and Structures, 24, No. 143, pp. 351–8 (1991).
11.65. J. TRITTHART, K. PETTERSSON and B. SORENSEN, Electrochemical removal of
chloride from hardened cement paste, Cement and Concrete Research, 23,
No. 5, pp. 1095–104 (1993).
11.66. J. TRITTHART. Concrete binding in cement. II. The influence of the hydroxide
concentration in the pore solution of hardened cement paste on chloride
binding, Cement and Concrete Research, 19, No. 5, pp. 683–91 (1989).
11.67. M.-J. AL-HUSSAINI et al., The effect of chloride ion source on the free chloride
ion percentages of OPC mortars, Cement and Concrete Research, 20, No. 5,
pp. 739–45 (1990).
11.68. L. TANG and L.-O. NILSSON, Chloride binding capacity and binding isotherms
of OPC pastes and mortars, Cement and Concrete Research, 23, No. 2, pp.
247–53 (1993).
11.69. RASHEEDUZZAFAR, A. S. AL-GAHTANI and S. S. AL-SAADOUN, Influence of
construction practices on concrete durability, ACI Materials Journal, 86, No.
6, pp. 566–75 (1989).
11.70. O. S. B. AL-AMOUDI et al., Prediction of long-term corrosion resistance of
plain and blended cement concretes, ACI Materials Journal, 90, No. 6, pp.
564–71 (1993).
11.71. S. NAGATAKI et al., Condensation of chloride ion in hardened cement matrix
materials and on embedded steel bars, ACI Materials Journal, 90, No. 4, pp.
323–32 (1993).
11.72. N. S. BERKE, Corrosion inhibitors in concrete, Concrete International, 13, No.
7, pp. 24–7 (1991).
11.73. C. K. NMAI, S. A. FARRINGTON and S. BOBROWSKI, Organic-based corrosion-
inhibiting admixture for reinforced concrete, Concrete International, 14, No.
4, pp. 45–51 (1992).
11.74. C. ALONSO and C. ANDRADE, Effect of nitrite as a corrosion inhibitor in
contaminated and chloride-free carbonated mortars, ACI Materials
Journal, 87, No. 2, pp. 130–7 (1990).
11.75. T. NIREKI and H. KABEYA, Monitoring and analysis of seawater salt
content, 4th Int. Conf. on Durability of Building Materials and Structures,
Singapore, pp. 531–6 (4–6 Nov. 1987).
11.76. W. H. HARRISON, Effect of chloride in mix ingredients on sulphate resistance
of concrete, Mag. Concr. Res., 42, No. 152, pp. 113–26 (1990).

628
 11.77. G. COCHET and B. JÉSUS, Diffusion of chloride ions in Portland cement–filler
mortars, Int. Conf. on Blended Cements in Construction, Sheffield UK, pp.
365–76 (Oxford, Elsevier Science, 1991).
11.78. A. J. AL-TAYYIB et al., Corrosion behavior of pre-rusted rebars after
placement in concrete, Cement and Concrete Research, 20, No. 6, pp. 955–
60 (1990).
11.79. B. MATHER, Calcium chloride in Type V-cement concrete, in Durability of
Concrete, ACI SP-131, pp. 169–76 (Detroit, Michigan, 1992).
11.80. D. W. S. Ho and R. K. LEWIS, The specification of concrete for reinforcement
protection – performance criteria and compliance by strength, Cement and
Concrete Research, 18, No. 4, pp. 584–94 (1988).
11.81. S. GOÑI, C. ANDRADE and C. L. PAGE, Corrosion behaviour of steel in high
alumina cement mortar samples: effect of chloride, Cement and Concrete
Research, 21, No. 4, pp. 635–46 (1991).
11.82. ACI 222R-89, Corrosion of metals in concrete, ACI Manual of Concrete
Practice Part 1: Materials and General Properties of Concrete, 30 pp.
(Detroit, Michigan, 1994).
11.83. A. LAMMKE, Chloride-absorption from concrete surfaces, in Evaluation and
Repair of Fire Damage to Concrete, ACI SP-92, pp. 197–209 (Detroit,
Michigan, 1986).
11.84. STRATEGIC HIGHWAY RESEARCH PROGRAM, SHRP-S-347, Chloride Removal
Implementation Guide, National Research Council, 45 pp. (Washington DC,
1993).
11.85. G. K. GLASS, C. L. PAGE and N. R. SHORT, Factors affecting the corrosion rate
of steel in carbonated mortars, Corrosion Science, 32, No. 12, pp. 1283–94
(1991).
11.86. STRATEGIC HIGHWAY RESEARCH PROGRAM SHRP-C-365, Mechanical Behavior of
High Performance Concretes, Vol. 5, National Research Council, 101 pp.
(Washington DC, 1993).
11.87. W. E. ELLIS JR., E. H. RIGG and W. B. BUTLER, Comparative results of
utilization of fly ash, silica fume and GGBFS in reducing the chloride
permeability of concrete, in Durability of Concrete, ACI SP-126, pp. 443–58
(Detroit, Michigan, 1991).
11.88. G. C. HOFF, Durability of offshore and marine concrete structures,
in Durability of Concrete, ACI SP-126, pp. 33–53 (Detroit, Michigan, 1991).
11.89. STUVO, Concrete in Hot Countries, Report of STUVO, Dutch member
group of FIP, 68 pp. (The Netherlands, 1986).
11.90. P. SCHIESSL and N. RAUPACH, Influence of blending agents on the rate of
corrosion of steel in concrete, in Durability of Concrete: Aspects of
Admixtures and Industrial By-products, 2nd International Seminar, Swedish
Council for Building Research, pp. 205–14 (June 1989).
11.91. R. F. M. BAKKER, Initiation period, in Corrosion of Steel in Concrete, Ed. P.
Schiessl, RILEM Report of Technical Committee 60-CSC, pp. 22–55
(London, Chapman and Hall, 1988).

629
 11.92. ACI 201.2R-92, Guide to durable concrete, ACI Manual of Concrete
Practice, Part 1: Materials and General Properties of Concrete, 41 pp.
(Detroit, Michigan, 1994).
11.93. Y. P. VIRMANI, Cost effective rigid concrete construction and rehabilitation in
adverse environments, Annual Progress Report, Year Ending Sept. 30, 1982,
U.S. Federal Highway Administration, 68 pp. (1982).
11.94. ACI 212.3R-91, Chemical admixtures for concrete, ACI Manual of Concrete
Practice, Part 1: Materials and General Properties of Concrete, 31 pp.
(Detroit, Michigan, 1994).
11.95. R. D. GAYNOR, Ready-mixed concrete, in Significance of Tests and Properties
of Concrete and Concrete-making Materials, Eds P. Klieger and J. F.
Lamond, ASTM Sp. Tech. Publ. No. 169C, pp. 511–21 (Philadelphia, Pa,
1994).
11.96. P. P. HUDEC, C. MACINNIS and M. MOUKWA, Microclimate of concrete barrier
walls: temperature, moisture and salt content, Cement and Concrete
Research, 16, No. 5, pp. 615–23 (1986).
11.97. O. E. GJØRV, K. TAN and M.-H. KHANG, Diffusivity of chlorides from seawater
into high-strength lightweight concrete, ACI Materials Journal, 91, No. 5,
pp. 447–52 (1994).
11.98. K. BYFORS, Influence of silica fume and flyash on chloride diffusion and pH
values in cement paste, Cement and Concrete Research, 17, No. 1, pp. 115–
30 (1987).
11.99. P. W. KEENE, Some tests on the durability of concrete mixes of similar
compressive strength, Cement Concr. Assoc. Tech. Rep. TRA/330 (London,
Jan. 1960).
11.100. E. SIEBEL, Air-void characteristics and freezing and thawing resistance of
superplasticized and air-entrained concrete with high workability,
in Superplasticizers and Other Chemical Admixtures in Concrete, Proc. 3rd
International Conference, Ottawa, Ed. V. M. Malhotra, ACI SP-119, pp.
297–320 (Detroit, Michigan, 1989).

630
Chapter 12. Testing of hardened concrete

We have seen that the properties of concrete are a function of time and ambient
humidity, and this is why, in order to be of value, tests on concrete have to be
performed under specified or known conditions. Different test methods and
techniques are used in different countries and sometimes even in the same country.
Because many of these tests are used in laboratory work, and especially in research, a
knowledge of the influence of the test methods on the measured property is of
importance. It is, of course, essential to distinguish between the effects of test
conditions and the intrinsic differences in the concretes being tested.
Tests can be made for different purposes but the main two objectives of testing are
quality control and compliance, now called conformity, with specifications.
Additional tests can be made for specific purposes, e.g. compressive strength tests to
determine the strength of concrete at transfer of prestress or at the time of the removal
of formwork. It should be remembered that tests are not an end in themselves: in
many practical cases, they do not lend themselves to a neat, concise interpretation, so
that, in order to be of real value, tests should always be used against the background
of experience. Nevertheless, because tests are generally performed for the purpose of
comparison with a specified, or some other, value, any departure from the standard
procedure is undesirable as it may lead to a dispute or to confusion.
Tests can be broadly classified into mechanical tests to destruction and non-
destructive tests which allow repeated testing of the same specimen and thus make
possible a study of the change in properties with time. Non-destructive tests also
permit testing concrete in an actual structure.
Tests for strength in compression
The most common of all tests on hardened concrete is the compressive strength test,
partly because it is an easy test to perform, and partly because many, though not all,
of the desirable characteristics of concrete are qualitatively related to its strength; but
mainly because of the intrinsic importance of the compressive strength of concrete in
structural design. Although invariably used in construction, the compressive strength
test has some disadvantages, but it has become, in French parlance,12.80 a part of the
engineer’s bagage culturel.
The strength test results may be affected by variation in: type of test specimen;
specimen size; type of mould; curing; preparation of the end surface; rigidity of the
testing machine; and rate of application of stress. For this reason, testing should
follow a single standard, with no departure from prescribed procedures.
Compressive strength tests on specimens treated in a standard manner which
includes full compaction and wet curing for a specified period give results
representing the potential quality of the concrete. Of course, the concrete in the
structure may actually be inferior, for example, due to inadequate compaction,
segregation, or poor curing. These effects are of importance if we want to know when
the formwork may be removed, or when further construction may continue, or the
structure be put into service. For this purpose, the test specimens are cured under
conditions as nearly similar as possible to those existing in the actual structure. Even
then, the effects of temperature and moisture would not be the same in a test specimen
as in a relatively large mass of concrete. The age at which service specimens are

631
tested is governed by the information required. On the other
hand, standard specimens are tested at prescribed ages, generally 28 days, with
additional tests often made at 3 and 7 days. Two types of compression test specimens
are used: cubes and cylinders. Cubes are used in Great Britain, Germany, and many
other countries in Europe. Cylinders are the standard specimens in the United States,
France, Canada, Australia, and New Zealand. In Scandinavia, tests are made on both
cubes and cylinders. The use of one or other type of specimen in a given country is so
ingrained that the European Standard BS EN 206:1996 allows the use of both
cylinders and cubes. Indeed, European standards use both types of strength.
Cube test
The specimens are cast in steel or cast-iron moulds of robust construction, generally
150 mm (or 6 in.) cubes, which should conform within narrow tolerances to the
cubical shape, prescribed dimensions and planeness. The mould and its base must be
clamped together during casting in order to prevent leakage of mortar. Before
assembling the mould, its mating surfaces should be covered with mineral oil, and a
thin layer of similar oil must be applied to the inside surfaces of the mould in order to
prevent the development of bond between the mould and the concrete.
The standard practice prescribed by BS EN 12390-1 : 2000 is to fill the mould in
one or more layers. Each layer of concrete is compacted by a vibrating hammer, or
using a vibrating table, or by a square steel punner. Ramming should continue until
full compaction without segregation or laitance has been achieved because it is
essential that the concrete in the cube be fully compacted if the test result is to be
representative of the properties of fully-compacted concrete. If, on the other hand, a
check on the properties of the concrete as placed is required, then the degree of
compaction of the concrete in the cube should simulate that of the concrete in the
structure. Thus, in the case of precast members compacted on a vibrating table, the
test cube and the member may be vibrated simultaneously, but the disparity of the two
masses makes the achievement of the same degree of compaction extremely difficult,
and this method is not recommended.
According to BS EN 12390-2 : 2009, after the top surface of the cube has been
finished by means of a float, the cube is stored undisturbed for 16 to 72 hours at a
temperature of 20 ± 5 °C (68 ± 9 °F) and a relative humidity preventing dehydration.
At the end of this period, the mould is stripped and the cube is further cured in water
or in a chamber with a relative humidity of not less than 95 per cent and at 20 ± 2 °C
(68 ± 4 °F).
In the compression test, the cube, while still wet, is placed with the cast faces in
contact with the platens of the testing machine, i.e. the position of the cube when
tested is at right angles to that as-cast. According to BS 1881-116 : 1983 (withdrawn),
the load on the cube should be applied at a constant rate of stress equal to 0.2 to 0.4
MPa/second (30 to 60 psi/second). ASTM C 39-09a prescribes a rate of 0.25 ± 0.05
MPa/sec (35 ± 7 psi/sec). Because of the non-linearity of the stress–strain relation of
concrete at high stresses, the rate of increase in strain must be increased progressively
as failure is approached, i.e. the speed of the movement of the head of the testing
machine has to be increased. The requirements for testing machines are discussed on
p. 590.
From the preceding discussion it can be seen that there is no unique simple relation
between the strengths of cubes and cylinders made from the same concrete. And yet,
European legislation, which made it possible for a contractor in any European Union

632
country to bid for, and build, concrete structures in any country, made it highly
desirable to describe the strength of concrete in a manner that allowed such
construction in an unequivocal manner. Consequently, European Standards chose to
assume that the strength of a cube is 5/4 of the strength of the cylinder and all mixes
therefore describe strength by a double number, e.g. 40/32, meaning that a cube
strength of 40 MPa is equivalent to a cylinder strength of 32 MPa. However, in
concrete made with lightweight aggregate, the ratio is much higher than 5/4.12.150
The compressive strength, known also as the crushing strength, is reported to the
nearest 0.5 MPa or 50 psi; a greater precision is usually only apparent.
Cylinder test
The standard cylinder is 6 in. in diameter, 12 in. long, or 150 by 300 mm, but in
France the size is 159.6 by 320 mm; the diameter of 159.6 mm gives a cross-sectional
area of 20 000 mm2. Cylinders are cast in a mould generally made of steel or cast iron,
with a clamped base; cylinder moulds are specified by ASTM C 470-09, which allows
also the use of single-use moulds, made of plastic, sheet metal and treated cardboard.
Details of moulds may seem to be trivial but non-standard moulds can result in a
misleading test result. For example, if the mould has a low rigidity, some of the
compaction effort is dissipated so that the compaction of the concrete in the mould
may be inadequate; a lower strength would be recorded. Conversely, if the mould
allows leakage of mix water, the strength of concrete would increase. Excessive re-
use of moulds intended for single use or for limited re-use leads to their distortion and
to an apparent loss of strength.12.55
The method of making test cylinders is prescribed by BS EN 12390-2 : 2009 and
by ASTM C 192-07. The procedure is similar to that used with cubes, but there are
differences in detail between the British and American standards.
The testing of a cylinder in compression requires that the top surface of the
cylinder be in contact with the platen of the testing machine. This surface,
when finished with a float, is not smooth enough for testing and requires further
preparation; this is a disadvantage of cylinders tested in compression. Treatment of
the top end of cylinders by capping is considered in a later section, but even though
the cylinders will be capped, ASTM C 192-07 and C 31-09 do not allow depressions
or excrescences greater than 3 mm ( in.); these could result in air pockets.12.55
Equivalent cube test
Sometimes, the compressive strength of concrete is determined using parts of a beam
tested in flexure. The end parts of such a beam are left intact after failure in flexure
and, because the beam is usually of square cross-section, an ‘equivalent’ or ‘modified’
cube can be obtained by applying the load through square steel plates of the same size
as the cross-section of the beam. It is important that the two plates be accurately
placed vertically above one another; a suitable jig is shown in Fig. 12.1. The specimen
should be placed so that the as-cast top surface of the beam is not in contact with
either plate.

633
 Fig. 12.1. Jig for testing equivalent cubes
The test is prescribed by BS 1881-119 : 1983 and ASTM C 116-90. The latter,
now withdrawn, allowed the use of beams whose cross-section is rectangular.
The strength of a modified cube is approximately the same as the strength of a
standard cube of the same size: actually, the restraint of the overhanging parts of the
‘cube’ may result in a slight increase in ultimate strength12.4 so that it is reasonable to
assume the strength of a modified cube to be, on average, 5 per cent higher than that
of a cast cube of the same size.
Effect of end condition of specimen and capping
When tested in compression, the top surface of the test cylinder is brought into
contact with the platen of the testing machine and, because this surface is not obtained
by casting against a machined plate but finished by means of a float, the top surface is
somewhat rough and not truly plane. Under such circumstances, stress concentrations
are introduced and the apparent strength of the concrete is reduced. Convex end
surfaces cause a greater reduction than concave ones as they generally lead to higher
stress concentrations. The reduction in the measured loss in strength is particularly
high in high strength concrete.12.5
To avoid this reduction in strength, plane end surfaces are essential: ASTM C 617-
09a requires the end surfaces of a cylinder to be plane within 0.05 mm (0.002 in.), as
determined by a straight edge and a feeler gauge, and to be perpendicular to the axis
of the cylinder within 0.5°. A method of testing concrete cylinders for planeness and
parallelism of end and perpendicularity of sides is prescribed in the U.S. Army Corps
Engineers Handbook for Concrete and Cement.12.81 While the procedures are not
unduly complex, such a test is most likely to be of interest in research work. A
limitation on planeness of the platens of the testing machine is prescribed by ASTM C
39-09a.
In addition to the absence of ‘high spots’, the contact surfaces should be free from
grains of sand or other debris (from a previous test), which would lead to premature
failure and, in extreme cases, to sudden splitting.

634
 There are three possible means of overcoming the ill-effects of an uneven end
surface of the specimen: capping, grinding, and packing with a bedding material.
Packing is not recommended because it results in an appreciable lowering of the
apparent mean strength of concrete, compared with capped, and often even with
smooth-trowelled, specimens (see Fig. 12.6). At the same time, the scatter of strength
results is appreciably reduced because the influence of the defects in planeness
(responsible for the large variation in strength) is eliminated.
The reduction in strength introduced by packing, usually of softboard, cardboard,
or lead, arises from lateral strains induced in the cylinder by the Poisson’s ratio effect
in the packing material. Poisson’s ratio of this material is generally higher than that of
concrete so that splitting is induced. This effect is similar to, although usually greater
than, that of lubricating the ends of the cylinder in order to eliminate the restraining
influence of the friction between the specimen and the platen on lateral spread of the
concrete. Such lubrication has been found to reduce the strength of the specimen.
Capping with a suitable material does not adversely affect the measured strength
and reduces its scatter compared with uncapped specimens. An ideal capping material
should have strength and elastic properties similar to those of the concrete in the
specimen; there is then no enhanced tendency to splitting, and a reasonably uniform
distribution of stress over the cross-section of the specimen is achieved.
The capping operation may be performed either just before testing or alternatively
soon after the specimen has been cast. Different materials are used in the two cases
but, whatever the capping material, it is essential that the cap be thin, preferably 1.5 to
3 mm ( to in.) thick. The capping material must be no weaker than the concrete in
the specimen; however, the strength of the cap is affected by its thickness. Too great a
difference in strength is thought to be undesirable because a very strong cap may
produce a large lateral restraint and thus lead to an apparent increase in strength. The
influence of the capping material on strength is much greater in the case of high- or
medium-strength concrete than in low-strength concrete;12.6,12.82 in the latter case,
Poisson’s ratio of the capping material is also of no influence. With 48 MPa (7000 psi)
concrete, high-strength capping leads to strengths 7 to 11 per cent higher than low-
strength capping. For 69 MPa (10 000 psi) concrete, the difference can be as high as
17 per cent. These differences are smaller when the thickness of the cap is very
small.12.82
Capping procedures are prescribed in ASTM C 617-09a. When the capping
operation is to be performed soon after casting, Portland cement paste is used. Before
its application, it is preferable to allow two to four hours’ delay after casting so that
the plastic shrinkage of the concrete and the resulting settlement of the top surface of
the material in the mould can take place. It is convenient to finish the original
concrete about 1.5 to 3 mm ( to in.) short of the top of the mould. During capping,
this space is filled with a stiff cement paste which has been allowed partially to shrink
and, by working down a glass or machined steel plate, a plane surface is obtained.
Experience is necessary to make this operation successful and particularly to obtain a
clean break between the cement paste and the plate: greasing the plate with a mixture
of lard oil and paraffin12.7 or covering with a thin film of graphite grease12.6 has been
found helpful. Following capping, moist curing must be continued.
The alternative method is to cap the cylinder shortly before it is tested: the actual
time depends on the hardening properties of the capping material. The cap should be 3

635
to 8 mm ( to in.) thick and it must bond well to the underlying concrete. Suitable
capping materials are high-strength gypsum plaster and molten sulfur mortar, but
regulated-set cement12.82 has also been used.
The sulfur mortar consists of sulfur and a granular material such as milled fire clay.
The mixture is applied in a molten state and allowed to harden with the specimen in a
jig which ensures a plane and square end surface. The use of a fume cupboard is
necessary because toxic fumes are produced. The sulfur mixture from tested cylinders
can be re-used up to five times, but care is required in selecting and using sulfur
mortar as otherwise the strength of the test cylinders can be significantly
affected.12.53 Moist curing must be resumed after capping.
An alternative to capping is to grind (using silicon carbide abrasion) the bearing
surface of the specimen until it is plane and square. This method produces very
satisfactory results but is rather expensive. It has been suggested that grinding leads to
a higher strength than capping in that any loss of strength associated with capping is
absent.12.84 Thus, ground specimens have the same strength as those with ‘perfect’ cast
test-surfaces.
Non-bonded caps
Although sulfur-mortar capping is satisfactory for concretes with strengths up to
about 100 MPa (or 14 000 psi), the capping operation is tedious and potentially
slightly dangerous. For this reason, a number of attempts have been made to develop
non-bonded caps. These are in the form of an elastomeric pad inserted into a
restraining rigid metal cap of the type shown in Fig. 12.2. Neoprene pads in steel caps
have been found to be satisfactory.12.74 The pad should fit snugly in the cap whose
internal diameter should be about 6 mm ( in.) larger than the diameter of the concrete
cylinder. It is important that the cylinder be concentric with the cap.

Fig. 12.2. Cross-section of a typical non-bonded capping system

The use of non-bonded rubber caps is permitted in Australia.12.75 It has been found
that the caps have to be fully moulded (and not punched) and that rubber of different
hardness has to be used according to the strength of concrete.12.75 This is a complicating
factor if the approximate strength of the cylinder cannot be anticipated. Furthermore,

636
rubber caps should not be used with low-strength concrete: limiting values are 20
MPa (or 3000 psi)12.75 and 30 MPa (or 4500 psi)12.73 have been suggested because, at
lower strengths, non-bonded caps lead to lower strength values than those obtained
with conventional sulfur-mortar capping.
The use of non-bonded caps has been limited in other countries and, therefore, a
reliable comparison of strengths obtained using these caps with values of strength of
sulfur-mortar capped cylinders is not available. However, even if there is a small
systematic difference in strength compared with the strength of sulfur-mortar capped
cylinders, this is not important because every method of capping introduces a
systematic influence on the observed strength so that there is no ‘true’ strength of
concrete. What is important is that a single method is used on a given construction
project.
The variability of test results on cylinders with non-bonded caps is smaller than
with standard caps. This may be due to the beneficial effect of non-bonded caps in
reducing the consequences of the roughness of cylinder ends.12.72
Capping very high strength concrete presents a special problem in that such
concrete has a higher strength than sulfur-mortar caps. Non-bonded caps are also
unsatisfactory because the pads can become seriously damaged and even extruded
from the cap.12.71 Grinding the cylinder ends gives very good results but it is slow and
expensive. Moreover, a high quality of grinding and lapping must be rigorously
ensured.
To avoid grinding, the use of a sand-filled restraining steel cap has been developed:
dry, fine siliceous sand is compacted in the cap; the cylinder is placed on top of the
sand; and molten paraffin is poured in order to form a seal confining the sand and
maintaining the centring of the cylinder.12.71 The compressive strengths of concretes up
to 120 MPa (or 17 000 psi) using sand capping agreed well with those of ground
specimens.12.71
For research purposes, the application of a truly uniform compressive stress may
be desirable. This has been achieved by loading through a mat of thin rubber strips
with gaps in-between,12.12 or through a stiff wire brush.12.56 A brush ‘platen’ consists of
filaments, about 5 by 3 mm (0.20 by 0.12 in.) in cross-section with gaps 0.2 mm
(0.008 in.) wide. This combination allows the free lateral deformation of concrete to
develop but the filaments do not buckle. The use of brush platens on 100 mm (4 in.)
cubes has been found to yield a strength equal to about 80 per cent of the strength
with rigid platens at a constant strain rate (for concrete strength of about 45 MPa
(6500 psi)).12.85
Testing of compression specimens
In addition to being plane, the end surfaces of the cylinder should be normal to its axis,
and this guarantees also that the end planes are parallel to one another. A small
tolerance is permitted: an inclination of the axis of the specimen to the axis of the
testing machine of 6 mm in 300 mm ( in. in 12 in.) has been found to cause no loss
of strength.12.5 The axis of the specimen when placed in the testing machine should be
as near to the axis of the platen as possible, but errors up to 6 mm ( in.) do not affect
the strength of cylinders made with low strength concrete.12.5 However, BS 1881-115 :
1986 (withdrawn) required a provision for positive and accurate location of the test
specimens. Likewise, a small lack of parallelism between the end surfaces of the

637
specimen does not adversely affect its strength, provided the testing machine is
equipped with a seating which can align freely with the end plane of the specimens.
The current British Standard for testing machines is BS EN 12390-4 : 2000.
Free alignment is achieved by a spherical seat. This can act not only when the
platens are brought into contact with the specimen but also when the load is being
applied. At this stage, some parts of the specimen may deform more than others. This
is the case in a cube in which, due to bleeding, the properties of different layers (as-
cast) may not be the same. In the testing position, the cube is at right angles to the as-
cast position so that the weaker and the stronger parts (parallel to one another) extend
from platen to platen. Under load, the weaker concrete, having a lower modulus of
elasticity, deforms more. With an effective spherical seat, the platen will follow the
deformation so that the stress on all parts of the cube is the same and failure occurs
when this stress reaches the strength of the weaker part of the cube. On the other hand,
if the platen does not change its inclination under load (i.e. moves parallel to itself) a
greater load is carried by the stronger part of the cube. The weaker part still fails first,
but the maximum load on the cube is reached only when the stronger part of the cube
carries its maximum load, too: thus the total load on the cube is greater than when the
platen is free to rotate. This behaviour was confirmed experimentally.12.9
To make the spherical seat of a testing machine effective under load, a highly polar
lubricant has to be used to reduce the coefficient of friction to a value as low as 0.04
(compared with 0.15 when a graphite lubricant is used).12.10 ASTM C 39-09a specifies
the use of petroleum-type oil such as conventional motor oil. It is not clear, however,
whether making this movement of the platen possible results in the observed strength
being more representative of the concrete under test. There are indications that a
machine with a platen that does not change inclination under load gives more
reproducible results when nominally similar cubes are tested.12.11 For this reason the
seat must not move under load. In any case, the observed strength is seriously affected
by the friction at the surface of the ball seat, so that, for tests to be comparable, it is
essential to maintain this surface in a standardized condition.
The loading of a platen through a spherical seat induces bending and distortion of
the platen, which depend on the thickness of the platen. ASTM C 39-09a prescribes
the thickness of the platen in relation to the size of the spherical seat, the latter being
governed by the size of the specimen.
Figure 12.3(a) indicates schematically the normal stress distribution at the platen-
concrete interface when a ‘hard’ platen is used: the compressive stress is then higher
near the perimeter than at the centre of the specimen. The same distribution exists
when the specimen or the platen are slightly concave. Conversely, when a ‘soft’
platen is used (Fig. 12.3(b)), the compressive stress is higher near the centre of the
specimen than around the perimeter. This condition is also produced by a slightly
convex specimen face or platen. In addition to the stress distributions of Fig. 12.3,
some local variations in stress exist due to the heterogeneity of concrete, and
specifically due to the presence of coarse aggregate particles near the end faces.

638
 Fig. 12.3. Normal stress distribution near ends of specimens when tested in a
machine with: (a) hard platens; (b) soft platens
A description of the different types of testing machines is outside the scope of this
book, but it ought to be mentioned that the failure of the specimen is affected by the
design of the machine, especially by the energy stored in it. With a very rigid machine,
the high deformation of the specimen under loads approaching the ultimate load is not
followed by the movement of the machine head, so that the rate at which the load is
applied decreases and a higher strength is recorded. On the other hand, in a less rigid
machine, the load follows more nearly the load-deformation curve for the specimen
and, when cracking commences, the energy stored by the machine is released rapidly.
This leads to failure under a lower load than would occur in a more rigid machine,
often accompanied by a violent explosion.12.8 The exact behaviour depends on the
detailed characteristics of the machine, not only its longitudinal stiffness, but also its
lateral stiffness being relevant.12.53 Proper and regular calibration of testing machines is
essential; this is prescribed in BS EN 12390-4 : 2000. The same standard gives a
method of verifying the performance of a testing machine.
Failure of compression specimens
On p. 293, we considered the failure of concrete subjected to uniaxial compression.
The compression test imposes, however, a rather more complex system of stress,
tangential forces being developed between the end surfaces of the concrete specimen
and the adjacent steel platens of the testing machine. In each material, the vertical
compression acting (the nominal stress on the specimen) results in a lateral expansion
due to the Poisson’s ratio effect. However, the modulus of elasticity of steel is 5 to 15
times greater, and Poisson’s ratio no more than twice greater, than the corresponding
values for concrete, so that the lateral strain in the platen is small compared with the
transverse expansion of the concrete if it were free to move. For instance, Newman
and Lachance12.57 found the lateral strain in a steel platen to be 0.4 of the lateral strain

639
in the concrete at a distance from the interface sufficient to remove the restraining
effect.
It can be seen then that the platen restrains the lateral expansion of the concrete in
the parts of the specimen near its ends: the degree of restraint exercised depends on
the friction actually developed. When the friction is eliminated, e.g. by applying a
layer of graphite or paraffin wax to the bearing surfaces, the specimen exhibits a large
lateral expansion and eventually splits along its full length.
With friction acting, i.e. under normal conditions of test, an element within the
specimen is subjected to a shearing stress as well as to compression. The magnitude
of the shearing stress decreases, and the lateral expansion increases, with an increase
in distance from the platen. As a result of the restraint, in a specimen tested to failure
there is a relatively undamaged cone or pyramid of height approximately equal to
(where d is the lateral dimension of the specimen).12.4 If the specimen is longer
than about 1.7d, a part of it will be free from the restraining effect of the platens. We
can note that specimens whose length is less than 1.5d show a considerably higher
strength than those with a greater length (see Fig. 12.5).
It seems then, that, when a shearing stress acts in addition to the uniaxial
compression, failure is delayed, and it can, therefore, be inferred that it is not the
principal compressive stress that induces cracking and failure but probably the lateral
tensile strain. The actual collapse may be due, at least in some cases, to the
disintegration of the core of the specimen. The lateral strain is induced by the
Poisson’s ratio effect and, assuming this ratio to be approximately 0.2, the lateral
strain is of the axial compressive strain. Now, we do not know the exact criteria of
failure of concrete but there are strong indications that failure occurs at a limiting
strain of 0.002 to 0.004 in compression or 0.0001 to 0.0002 in tension. Because the
ratio of the latter of these strains to the former is less than Poisson’s ratio of concrete,
it follows that conditions of failure in circumferential tension are achieved before the
limiting compressive strain has been reached.
Vertical splitting has been observed in numerous tests on cylinders, particularly in
high-strength specimens made of mortar or neat cement paste, and also in sulfur-
infiltrated concrete. The effect is less common in ordinary concrete when coarse
aggregate is present because it provides lateral continuity.12.4 The presence of vertical
cracks has also been confirmed by measurements of ultrasonic pulse velocity along
and across the specimen.12.13
The observations on the actual stress distribution in a nominally uniaxial situation
do not necessarily detract from the value of the compression test as a comparative test,
but we should be wary of interpreting it as a true measure of the compressive strength
of concrete.
Effect of height/diameter ratio on strength of cylinders
Standard cylinders are of height h equal to twice the diameter d, but sometimes
specimens of other proportions are encountered. This is particularly the case with
cores cut from in situ concrete: the diameter depends on the size of the core-cutting
tool whereas the height of the core varies with the thickness of the slab or member. If
the core is too long, it can be trimmed to the h/d ratio of 2 before testing but, with too
short a core, it is necessary to estimate the strength of the same concrete as if it had
been determined on a specimen with h/d = 2.

640
 ASTM C 42-04 and BS EN 12504-1 : 2009 give correction factors (Table 12.1) but
Murdock and Kesler12.14 found that the correction depends also on the level of strength
of the concrete (Fig. 12.4). High strength concrete is less affected by the
height/diameter ratio of the specimen, and such a concrete is also less influenced by
the shape of the specimen; the two factors should be related as there is comparatively
little difference between the strengths of a cube and of a cylinder with h/d = 1.

Table 12.1. Standard Correction Factors for Strength of Cylinders with Different
Ratios of Height to Diameter

Fig. 12.4. Influence of the height/diameter ratio on the apparent strength of a

cylinder for different strength levels12.14

641
 The influence of strength on the conversion factor is of practical significance in the
case of low strength concrete, if cores with h/d smaller than 2 are tested. Using ASTM
C 42-04 and, even more so, BS EN 12504-1 : 2009, factors the strength that would be
obtained with an h/d ratio of 2 would be overestimated; yet, it is in the case of
concrete of low strength, or suspected of having too low a strength, that a correct
estimate of strength is often particularly important.
The general pattern of influence of h/d on the strength of low- and medium-
strength concrete is shown in Fig. 12.5. For values of h/d smaller than 1.5
the measured strength increases rapidly due to the restraining effect of the platens of
the testing machine. When h/d varies between about 1.5 and 4, strength is affected
only little and, for h/d values between 1.5 and 2.5, the strength is within 5 per cent of
the strength of standard specimens (h/d = 2). For values of h/d above 5, strength falls
off more rapidly, the effect of the slenderness ratio becoming apparent.

Fig. 12.5. General pattern of influence of the height/diameter ratio on the

apparent strength of a cylinder12.40
It seems thus that the choice of the standard height/diameter ratio of 2 is suitable,
not only because the end effect is largely eliminated and a zone of uniaxial
compression exists within the specimen, but also because a slight departure from this

642
ratio does not seriously affect the measured value of strength. No correction is
required for values of h/d between 1.94 and 2.10.
The influence on strength of the ratio of height to the least lateral dimension
applies also in the case of prisms.
Of course, if the end friction is eliminated, the effect of h/d on strength disappears
but this is very difficult to achieve in a routine test. The general pattern of the
influence of packing between the platen and the specimen on the strength of cylinders
with different values of h/d is shown in Fig. 12.6.

Fig. 12.6. Relative strength of cylinders of different height/diameter ratios with

various types of packing between the platens and the specimen.12.58 (Strength of
cylinder with h/d = 2 and no packing taken as 1.0): (A) no packing; (B) 8 mm
( in.) soft wallboard; (C) 25 mm (1 in.) plastic board
The end effect decreases more rapidly the more homogeneous the material; it is
thus less noticeable in mortars and probably also in lightweight aggregate concrete of
low or moderate strength where a lower heterogeneity arises from the smaller
difference between the elastic moduli of the cement paste and the aggregate than is
the case with normal weight aggregate. It has been found that, with lightweight
aggregate concrete, the value of the ratio of strengths of a standard cylinder to a
cylinder with a height-diameter ratio of 1 is between 0.95 and 0.97.12.15,12.60 This has,
however, not been confirmed in Russian tests on concrete made with expanded clay
aggregate where a ratio of about 0.77 was reported.12.59
Comparison of strengths of cubes and cylinders
We have seen that the restraining effect of the platens of the testing machine extends
over the entire height of a cube but leaves unaffected a part of a test cylinder. It is,

643
therefore, to be expected that the strengths of cubes and cylinders made from the same
concrete differ from one another.
According to the expressions converting the strength of cores into the strength of
equivalent cubes in BS 1881 : 1983 (withdrawn), the strength of cylinder is equal to
0.8 of the strength of a cube but, in reality, there is no simple relation between the
strengths of the specimens of the two shapes. The ratio of the strengths of the cylinder
to the cube increases strongly with an increase in strength12.16 and is nearly 1 at
strengths of more than 100 MPa (or 14 000 psi). Some other factors, for example, the
moisture condition of the specimen at the time of testing, have also been found to
affect the ratio of strengths of the two types of specimens.
Because European Standard BS EN 206-1 : 2000 recognizes the use of both
cylinders and cubes, it includes a table of equivalence of strengths of the two types of
compression specimens up to 50 MPa (measured on cylinders). The values of the
cylinder/cube strength ratio are all around 0.8. The CEB–FIP Design Code12.1 gives a
similar table of equivalence but, above 50 MPa, the cylinder/cube strength ratio rises
progressively, reaching 0.89 when the cylinder strength is 80 MPa. Neither of these
tables should be used for purposes of conversion of a measured strength of one type
of specimen to the strength of the other type. For any one construction project, a
single type of compressive strength test specimen should be used.
It is difficult to say which type of specimen, cylinder or cube, is ‘better’ but, even
in countries where cubes are the standard specimen, there seems to be a tendency, at
least for research purposes, to use cylinders rather than cubes, and this has been
recommended by RILEM (Réunion Internationale des Laboratoires d’Essais et de
Recherches sur les Matériaux et les Constructions) – an international organization of
testing laboratories. Cylinders are believed to give a greater uniformity of results for
nominally similar specimens because their failure is less affected by the end restraint
of the specimen; their strength is less influenced by the properties of the coarse
aggregate used in the mix; and the stress distribution on horizontal planes in a
cylinder is more uniform than on a specimen of square cross-section.
It may be recalled that cylinders are cast and tested in the same position, whereas
in a cube the line of action of the load is at right angles to the axis of the cube as-cast.
In structural compression members, the situation is similar to that existing in a test
cylinder, and it has been suggested that, for this reason, tests on cylinders are more
realistic. The relation between the directions as-cast and as-tested has, however, been
shown not to affect appreciably the strength of cubes made with unsegregated and
homogeneous concrete12.3 (Fig. 12.7). Moreover, as shown earlier, the stress
distribution in any compression test is such that the test is only comparative and offers
no quantitative data on the strength of a structural member.

644
 Fig. 12.7. Relation between mean strength of concrete cubes loaded in the
direction of casting and in the standard manner12.3
Tests for strength in tension
Although concrete is not normally designed to resist direct tension, the knowledge of
tensile strength is of value in estimating the load under which cracking will develop.
The absence of cracking is of considerable importance in maintaining the continuity
of a concrete structure and in many cases in the prevention of corrosion of
reinforcement. Cracking problems occur when diagonal tension arising from shearing
stresses develops, but the most frequent case of cracking is due to restrained shrinkage
and temperature gradients. An appreciation of the tensile strength of concrete helps in
understanding the behaviour of reinforced concrete even though the actual design
calculations do not in many cases explicitly take the tensile strength into account. The
rather extensive topic of cracking is considered in Chapter 10.
Strength in tension is of interest also in unreinforced concrete structures, such as
dams, under earthquake conditions. Other structures, such as highway and airfield
pavements, are designed on the basis of flexural strength, which involves strength in
tension.
There are three types of test for strength in tension: direct tension test, flexure test,
and splitting tension test.
A direct application of a pure tension force, free from eccentricity, is very difficult.
Despite some success with the use of lazy-tong grips,12.19 it is difficult to avoid

645
secondary stresses such as those induced by grips or by embedded studs. A direct
tension test, using bonded end plates, is prescribed by the U.S. Bureau of
Reclamation.12.17 The other two types of test for strength in tension are considered
below.
Flexural strength tests
In these tests, a plain (unreinforced) concrete beam is subjected to flexure using
symmetrical two-point loading until failure occurs. Because the load points are spaced
at one-third of the span, the test is called a third-point loading test. The theoretical
maximum tensile stress reached in the bottom fibre of the test beam is known as
the modulus of rupture.
Beams are normally tested on their side in relation to the as-cast position but,
provided the concrete is unsegregated, the position of the beam as tested relative to
the as-cast position does not affect the modulus of rupture.12.22,12.23
British Standard BS EN 12390-5 : 2000 prescribes third-point loading on 150 by
150 by 750 mm (6 by 6 by 30 in.) beams supported over a span of 450 mm (18 in.)
but 100 by 100 mm (4 by 4 in.) beams can also be used, provided the beam side is at
least three times the maximum size of the aggregate.
The requirements of ASTM C 78-09 are similar to those of BS EN 12390-5 : 2009.
If fracture occurs within the central one-third of the beam, the modulus of rupture is
calculated on the basis of ordinary elastic theory, and is thus equal to:
PL/(bd2),
where P = maximum total load on the beam
L = span
b = width of the beam, and
d = depth of the beam.
If, however, fracture occurs outside the load points, say, at a distance a from the
near support, a being the average distance measured on the tension surface of the
beam, but not more than 5 per cent of the span, then the modulus of rupture is given
by 3Pa/(bd2). This means that the maximum tensile stress at the critical section, and
not the maximum stress on the beam, is considered in the calculations. The British
approach is to disregard failure outside the middle-third of the beam.
There exists also a test for flexural strength under centre-point loading, prescribed
in ASTM C 293-08, and also in British Standard BS EN 12390-5 : 2000. In this test,
failure occurs when the tensile strength of concrete in the extreme fibre immediately
under the load point is exhausted. On the other hand, under third-point loading, one-
third of the length of the extreme fibre in the beam is subjected to the maximum stress,
so that the critical crack may develop at any section in one-third of the beam length.
Because the probability of a weak element (of any specified strength) being subjected
to the critical stress is considerably greater under two-point loading than when a
central load acts, the centre-point loading test gives a higher value of the modulus of
rupture,12.20 but also a more variable one. In consequence, the centre-point loading test
is very rarely used.
The expression for the modulus of rupture, given earlier in this section, was
qualified by the term ‘theoretical’ because it is based on the elastic beam theory, in
which the stress–strain relation is assumed to be linear, so that the tensile stress in the
beam is assumed to be proportional to the distance from its neutral axis. In fact, as

646
discussed in Chapter 9, there is a gradual increase in strain with an increase in stress
above about one-half of the tensile strength. In consequence, the shape of the actual
stress block under loads nearing failure is parabolic, and not triangular. The modulus
of rupture thus overestimates the tensile strength of concrete: Raphael12.52 showed that
the correct value of tensile strength is about of the theoretical modulus of rupture
(see Fig. 12.8).

Fig. 12.8. Plot of the splitting-tension strength and of of the modulus of rupture
against compressive strength of concrete (based on ref. 12.52)
There exist additional possible reasons why the modulus of rupture test gives a
higher value of strength than a direct tension test made on the same concrete. First,
any accidental eccentricity in a direct tension test results in a lower apparent strength
of the concrete. The second argument is similar to that justifying the influence of the
loading arrangement on the value of the modulus of rupture: under direct tension, the
entire volume of the specimen is subjected to the maximum stress, so that the
probability of a weak element occurring is high. Third, in the flexure test, the
maximum fibre stress reached may be higher than in direct tension because the

647
propagation of a crack is blocked by less-stressed material nearer to the neutral axis.
Thus, the energy available is below that necessary for the formation of new crack
surfaces. These various reasons for the difference between the modulus of rupture and
the direct tensile strength are not all of equal importance.
At the beginning of this chapter, it was mentioned that flexural strength of concrete
is of interest in the design of pavement slabs. However, the flexure test is not
convenient for control or compliance purposes because the test specimens are heavy
and are easily damaged. Also, the outcome of the flexure test is strongly affected by
the moisture conditions of the specimen (see p. 602) and, more generally, the
variability of the modulus of rupture is large.12.115 It is, therefore, convenient to
establish experimentally a relation between the modulus of rupture and cylinder
compressive strength, and to use the latter in routine testing.12.2 The relation between
tensile and compressive strengths is discussed on p. 310.
A wide range of tests12.131 has shown a linear relation between the modulus of
rupture and the splitting tensile strength at a given age. This finding is of value if the
strength of pavement concrete in situ has to be determined: cutting cores as well as
testing them, in compression or in splitting tension, is very much easier than cutting
out beams for the modulus of rupture test. Moreover, cores are frequently cut anyway
for the purpose of verifying the thickness of pavement.
Splitting tension test
In this test, a concrete cylinder, of the type used for compression tests, is placed with
its axis horizontal between the platens of a testing machine, and the load is increased
until failure by indirect tension in the form of splitting along the vertical diameter
takes place.
If the load is applied along the generatrix, then an element on the vertical diameter
of the cylinder (Fig. 12.9) is subjected to a vertical compressive stress of:

and a horizontal tensile stress of 2P/(πLD)

where P = compressive load on the cylinder
L = length of the cylinder
D = diameter, and
r and (D – r) = distances of the element from the two loads respectively.

648
 Fig. 12.9. The splitting test
However, immediately under the load, a high compressive stress would be induced
and, in practice, narrow strips of a packing material, such as plywood, are interposed
between the cylinder and the platens. Without packing strips, the recorded strength is
lower, typically by 8 per cent. ASTM C 496-04 prescribes plywood strips, 3 mm
( in.) thick and 25 mm (1 in.) wide. British Standard BS EN 12390-6 : 2009 specifies
hardboard strips, 4 mm thick and 15 mm wide. With such an arrangement, the
distribution of the horizontal stress on a section containing the vertical diameter is as
shown in Fig. 12.10. The stress is expressed in terms of 2P/(πLD), and it can be seen
that a high horizontal compressive stress exists in the vicinity of the loads but, as this
is accompanied by a vertical compressive stress of comparable magnitude, thus
producing a state of biaxial stress, failure in compression does not take place.

649
 Fig. 12.10. Distribution of horizontal stress in a cylinder loaded over a width
equal to of the diameter12.24 (Crown copyright)
During the splitting test, the platens of the testing machine should not be allowed
to rotate in a plane perpendicular to the axis of the cylinder, but a slight movement in
the vertical plane containing the axis should be permitted in order to accommodate a
possible non-parallelism of the generatrices of the cylinder. This can be achieved by
means of a simple roller arrangement interposed between one platen and the cylinder.
The rate of loading is prescribed by ASTM C 496-04 and by BS EN 12390-6 : 2009.
Cubes and prisms can also be subjected to the splitting test, the load being applied
through loading pieces resting against the cube on centre lines of two opposing faces.
The cube test, covered by BS 1881 : 117 : 1983 (replaced by BS EN 12390-6 : 2009),
gives the same result as the splitting test on a cylinder,12.144 viz. the horizontal tensile
stress is equal to 2P/(πa2) where a is the side of the cube. This means that only the
concrete within a cylinder inscribed in the cube resists the applied load.
An advantage of the splitting test is that the same type of specimen can be used for
both the compression and the tension tests. Therefore, the splitting cube test is of
interest only in countries where the cube and not the cylinder is used as a standard
compression specimen; few data are available on the performance of the splitting cube
test.
The splitting test is simple to perform and gives more uniform results than other
tension tests.12.24 The strength determined in the splitting test is believed to be close to
the direct tensile strength of concrete, being 5 to 12 per cent higher. It has been
suggested, however, that, in the case of mortar and lightweight aggregate concrete, the
splitting test yields too low a result. With normal aggregate, the presence of large
particles near the surface to which the load is applied may influence the behaviour.12.86
It may be noted that, according to ACI 318-0812.124 splitting tensile strength should
not be used for the purpose of establishing conformity.

650
Influence on strength of moisture condition during test
The British as well as ASTM Standards require that all the test specimens be tested in
a ‘wet’ or ‘moist’ condition. This condition has the advantage of being better
reproducible than a ‘dry condition’ which includes widely varying degrees of dryness.
Occasionally, a test specimen may not be in a wet condition, and it is of interest to
consider what are the consequences of such departure from the standard. It should be
emphasized that only the condition immediately prior to the test is considered, it being
assumed that usual curing has been applied in all cases.
As far as compressive strength specimens are concerned, testing in a dry condition
leads to a higher strength. It has been suggested12.51 that drying shrinkage at the surface
induces a biaxial compression on the core of the specimen, thus increasing its strength
in the third direction, that is, in the direction of the applied load. However, tests have
shown that well-cured mortar prisms12.50 and concrete cores,12.121 when completely dried,
had a higher compressive strength than when tested wet. These specimens were not
subject to differential shrinkage so that there was no biaxial stress system induced.
The behaviour of the specimens, described above, accords also with the
suggestion12.32 that the loss of strength due to wetting of a compression test specimen is
caused by the dilation of the cement gel by adsorbed water: the forces of cohesion of
the solid particles are then decreased. Conversely, when on drying the wedge-action
of water ceases, an apparent increase in strength of the specimen is recorded. The
effects of water are not merely superficial as dipping the specimens in water has much
less influence on strength than soaking. On the other hand, soaking concrete in
benzene or paraffin, known not to be adsorbed by the cement gel, has no influence on
strength. Re-soaking oven-dried specimens in water reduces their strength to the value
of continuously wet-cured specimens, provided they have hydrated to the same
degree.12.32 The variation in strength due to drying appears thus to be a reversible
phenomenon.
The quantitative influence of drying varies: with 34 MPa (5000 psi) concrete, an
increase in compressive strength up to 10 per cent has been reported12.33 on thorough
drying, but if the drying period is less than 6 hours, the increase is generally less than
5 per cent. Other tests have shown the decrease in strength, in consequence of 48-hour
wetting prior to test, to be between 9 and 21 per cent.12.49
Beam specimens tested in flexure exhibit a behaviour opposite to that of
compression test specimens: a beam which has been allowed to dry before testing has
a lower modulus of rupture than a similar specimen tested in a wet condition.12.109 This
difference is due to the tensile stresses induced by restrained shrinkage prior to the
application of the load which induces tension in the extreme fibre. The magnitude of
the apparent loss of strength depends on the rate at which moisture evaporates from
the surface of the specimen. It should be emphasized that this effect is distinct from
the influence of curing on strength.
If, however, the test specimen is small and drying takes place very slowly, so that
internal stresses can be redistributed and alleviated by creep, an increase in strength is
observed. This was found in tests on concrete beams,12.31 and also on mortar
briquettes.12.30 Conversely, wetting a completely dry specimen prior to testing reduces
its strength;12.31 interpretation of this phenomenon is controversial.12.128
The strength of cylinders tested in splitting tension is not affected by the moisture
condition because failure occurs in a plane remote from the surface subjected to
wetting or drying.

651
 The temperature of the specimen at the time of testing (as distinct from the curing
temperature) affects the strength, a higher temperature leading to a lower indicated
strength, both in the case of compression and of flexure specimens (Fig. 12.11).

Fig. 12.11. Influence of temperature at the time of testing on strength

Influence of size of specimen on strength
The size of test specimens for strength testing is prescribed in the relevant standards,
but occasionally more than one size is permitted. Moreover, from time to time
arguments in favour of use of smaller specimens are advanced. These point out their
advantages: smaller specimens are easier to handle and are less likely to be
accidentally damaged; the moulds are cheaper; a lower capacity testing machine is
needed; and less concrete is used, which in the laboratory means less storage and
curing space, and also a smaller quantity of aggregate to be processed.12.41 On the other
hand, the size of the test specimen may affect the resulting strength and also the
variability of test results. For these reasons, it is important to consider in detail the
influence of the size of specimen on strength test results.
The discussion on p. 292 showed that concrete is composed of elements of variable
strength so that it is reasonable to assume that the larger the volume of the concrete
subjected to stress the more likely it is to contain an element of a given extreme (low)
strength. As a result, the measured strength of a specimen decreases with an increase
in its size, and so does the variability in strength of geometrically similar specimens.
Because the influence of size on strength depends on the standard deviation of
strength (Fig. 12.12) it follows that the size effects are smaller the greater the
homogeneity of the concrete. Thus, the size effect in lightweight aggregate concrete
should be smaller, but this has not been confirmed with any degree of certainty,
although there is some support for this suggestion in the available data.12.76 Figure
12.12 can also explain why the size effect virtually disappears beyond a certain size of

652
the specimen: for each successive ten fold increase in size of the specimen it loses
progressively a smaller amount of strength.

Fig. 12.12. Strength distribution in samples of size n for an underlying normal

distribution12.34
On p. 292, the concept of the weakest link was discussed; to use this concept we
require the knowledge of the distribution of extreme values in samples of a given size,
drawn at random from a parent population with a given distribution of strength. This
distribution is generally not known, and certain assumptions regarding its form have
to be made. Here, it will suffice to give Tippett’s12.34 data on the variation in strength
and standard deviation of samples of size n in terms of the strength and standard
deviation of a sample of unit size, when the unit sample has a normal distribution of
strength. Figure 12.12 shows this variation in strength for samples when n equals 10,
102, 103, and 105.
In the case of tests on the strength of concrete, we are interested in the averages of
extremes as a function of the size of the specimen. Average values of samples chosen
at random tend to have a normal distribution, so that the assumption of this type of
distribution, when average values of samples are used, does not introduce serious
error, and has the advantage of simplifying the computations. In some practical cases,
a skewness of distribution has been observed; this may not be due to any ‘natural’
properties of concrete but to the rejection of poor quality concrete on the site so that

653
such concrete never reaches the testing stage.12.35 A full treatment of the statistical
aspects of testing is outside the scope of this book.*
*
See J. B. KENNEDY and A. M. NEVILLE, Basic Statistical Methods for Engineers and
Scientists, 3rd Ed. (New York and London, Harper and Row, 1986).
Size effects in tensile strength tests
Figure 12.12 shows that both the mean strength and dispersion decrease with an
increase in the size of the specimen. Experimental results have confirmed this pattern
of behaviour in modulus of rupture tests12.20,12.23 (see Figs 12.13 and 12.14) and also in
direct tension12.19 and in indirect tension.12.64

Fig. 12.13. Modulus of rupture of beams of different sizes subjected to centre-

point and third-point loading12.20 (Crown copyright)

654
 Fig. 12.14. Coefficient of variation of the modulus of rupture for beams of
different sizes12.23
Direct tension tests on cylinders of concretes with compressive strengths between
35 and 128 MPa (5000 and 18 500 psi) were performed by Rossi et al.12.97 They
confirmed the decrease in tensile strength and also in variability of test results with an
increase in size: the decrease in strength is larger the lower the strength of concrete
(see Fig. 12.15). The coefficient of variation also decreases with an increase in size of
the specimen, as shown in Fig. 12.16, but there is no apparent effect of the strength of
concrete on this relation. Rossi et al.12.97 explain this influence of strength in terms of
the heterogeneity of the mix components. Specifically, the size effect is a function of
the ratio of the specimen size to the maximum size of aggregate and of the difference
in strength between the aggregate particles and the surrounding mortar. This
difference is small in very high strength concrete and also in lightweight aggregate
concrete.12.97

655
 Fig. 12.15. Direct tensile strength of concrete cylinders tested by Rossi et
al.12.97 plotted as a function of cylinder diameter

656
Fig. 12.16. Coefficient of variation of direct tensile strength of cylinders tested by
Rossi et al.12.97 plotted as a function of cylinder diameter
Splitting tension tests on 150 mm diameter by 300 mm high (6 by 12 in.) cylinders
and 100 mm diameter by 200 mm high (4 by 8 in.) cylinders have given an average
ratio12.131 of the strength of the former to the latter of 0.87; the average splitting tension
strength of the larger cylinders was 2.9 MPa (415 psi). The standard deviation for the
larger cylinders was 0.18 MPa (26 psi) and, for the smaller, 0.27 MPa (39 psi). The
coefficients of variation were, respectively, 6.2 and 8.2 per cent. It is worth observing
that the coefficient of variation of the splitting tension strength of 150 by 300 mm
cylinders had nearly the same value as the coefficient of variation of the modulus of
rupture determined on beams with a 150 by 150 mm (6 by 6 in.) cross-section made
of the same concrete.12.131

657
 The influence of the cylinder size on splitting tension strength was confirmed by
Bažant et al.12.94 on the basis both of their own tests on mortar discs and also on the
basis of tests on concrete cylinders performed by Hasegawa et al. In both these series
of tests, the size effect disappears in large-size specimens; this topic is discussed in
the next section.
Cement compacts have also been found to show the size effect when tested in
splitting tension.12.93 The same applies in the case of the ring test.12.64
Size effects in compressive strength tests
Let us now consider size effects in compressive strength specimens. Figure
12.17 shows the relation between mean strength and specimen size for cubes,
and Table 12.2 gives the relevant values for standard deviation.12.18 Prisms12.36,12.37 and
cylinders12.38 exhibit a similar behaviour (Fig. 12.18). The size effects are, of course,
not limited to concrete, and have been found also in anhydrite12.39 and other materials.

Fig. 12.17. Compressive strength of cubes of different sizes12.35

Table 12.2. Standard Deviation of Cubes of Different sizes12.18

658
 Fig. 12.18. Compressive strength of cylinders of different sizes12.38
It is interesting to note that the size effect disappears beyond a certain size so that a
further increase in the size of a member does not lead to a decrease in strength, both in
compression12.38 and in splitting tension.12.94 According to the U.S. Bureau of
Reclamation,12.77 the strength curve becomes parallel to the size axis at a diameter of
457 mm (18 in.), i.e. cylinders of 457 mm (18 in.), 610 mm (24 in.), and 914 mm (36
in.) diameter all have the same strength. The same investigation indicates that the
decrease in strength with an increase in size of the specimen is less pronounced in
lean mixes than in rich ones. For instance, the strength of 457 mm (18 in.) and 610
mm (24 in.) cylinders relative to 152 mm (6 in.) cylinders is 85 per cent for rich mixes
but 93 per cent for lean (167 kg/m3 (282 lb/yd3)) mixes (cf. Fig. 12.18).

659
 These experimental data are of importance in refuting a speculation that, if the size
effect is extrapolated to very large structures, a dangerously low strength might be
expected. Evidently this is not so because local failure is not tantamount to collapse.
The various test results on the size effect are of interest because size effects have
been ascribed to a variety of causes: the wall effect; the ratio of the specimen size to
the maximum aggregate size; the internal stresses caused by the difference in
temperature and humidity between the surface and the interior of the specimen; the
tangential stress at the contact surface between the platen of the testing machine and
the specimen due to friction or bending of the platen; and the difference in the
effectiveness of curing. The last suggestion, for instance, is disproved by
Gonnerman’s12.40 results (Fig. 12.19) which show that specimens of different size and
shape gain strength at the same rate. In this connection, Day and Haque12.90 showed that
the relation between the strength of 150 by 300 mm (6 by 12 in.) and 75 by 150 mm
(3 by 6 in.) cylinders is not affected by the method of curing.

Fig. 12.19. Effect of age on the compressive strength of specimens of different

shape and size12.40 (mix 1:5 by volume)

660
 Within the range of sizes of specimens normally used, the effect of size on strength
is not large, but it is significant and should not be ignored in work of high accuracy or
in research. Analysis of numerous test data12.65 has suggested a general relation
between the compressive strength of concrete and the shape and size of the specimen
in terms of V/hd + h/d, where V = volume of specimen, h = its height, and d = its least
lateral dimension. Figure 12.20 indicates the fit of experimental data to the relation
postulated. The validity of the form of the relation in high strength concrete has been
confirmed.12.148

Fig. 12.20. General relation between ratio of strength of concrete specimens fc to

strength of 6 in. cube fcu,6, and V/6hd + h/d, where V is volume of specimen, h its
height and d is its least lateral dimension. (All dimensions in inches; in millimetre
units, fcu,6 would become fcu,152 and on the right-hand side, 6 would be replaced by
152)
In direct tension, strength was found to be proportional to Vn, where n varies
between –0.02 and –0.04, depending on the type of aggregate.12.91 Thus, if cylinders
150 mm (6 in.) in diameter have a strength of 1.0, 50 mm (2 in.) cylinders would have
a strength of 1.05 to 1.08, and 200 mm (8 in.) cylinders 0.97 to 0.99. Prisms were
found to have similar behaviour. It was also found that the coefficient of variation
decreases as the specimen size increases.12.91 Torrent12.92 confirmed that the volume of
‘highly stressed’ concrete directly influences the strength of concrete in various
tension tests; this description was used to denote concrete stressed to about 95 per
cent of the maximum stress. Torrent’s expression involved the term Vn, but, in his
tests, n appeared to be independent of the type of aggregate or the water/cement ratio.

661
 The discussion in this section shows that, within the range of usual specimen sizes,
the influence of size on the average strength is not large for most practical purposes.
However, because of the higher scatter of results obtained with smaller specimens,
they have to be used in a greater number to give the same precision of the mean: five
to six 100 mm (4 in.) concrete cubes would be required instead of three 150 mm (6 in.)
cubes;12.42 or five 13 mm ( in.) mortar cubes instead of two 100 mm (4 in.) cubes.12.43
If the usual sets of three compressive strength cylinders are used, then, by
changing from 150 by 300 mm (6 by 12 in.) cylinders to 75 by 150 mm (3 by 6 in.)
cylinders, the coefficient of variation of the 28-day strength would typically rise from
3.7 to 8.5 per cent.12.88 Such an increase in the variability is a serious drawback to the
use of smaller test specimens.
Specimen size and aggregate size
If the maximum size of aggregate is large in relation to the size of the mould, the
compaction of concrete and the uniformity of distribution of the large particles of
aggregate are affected. This is known as the wall effect because the wall influences
the packing of the concrete: the quantity of mortar required to fill the space between
the particles of the coarse aggregate and the wall is greater than that necessary in the
interior of the mass, and therefore greater than the quantity of mortar available in a
well-proportioned mix (Fig. 12.21). In tests on concrete made with 19.05 mm ( in.)
aggregate, 101.6 mm (4 in.) cubes have been found to require for full compaction an
increase in sand content equal to 10 per cent of the total mass of aggregate, compared
with a mix used in an infinitely large section.12.44 To make up this deficiency of fine
material during the actual making of specimens, mortar would need to be added from
the remainder of the mix.

Fig. 12.21. ‘Wall effect’

The wall effect is more pronounced the larger the surface/volume ratio of the
specimen and is, therefore, smaller in flexure test specimens than in cubes or
cylinders.
To minimize the wall effect, various standards specify the minimum size of the test
specimen in relation to the maximum size of aggregate. British Standards BS 1881-

662
108 : 1983 and BS 1881-110 : 1983 (both withdrawn), respectively, allow the use of
100 mm cubes and 100 by 200 mm cylinders with aggregates whose maximum size is
up to 20 mm; 150 mm cubes and 150 by 300 mm cylinders can be used with
aggregate up to 40 mm in size. The requirement of ASTM C 192-07 is that the
diameter of the test cylinder or the minimum dimension of a prism be at least 3 times
the nominal maximum size of aggregate.
When the aggregate size exceeds the permissible value for the mould used,
screening out of the large-size aggregate is sometimes resorted to. This operation is
called wet screening. The screening must be done quickly in order to avoid drying out,
and the screened material should be remixed by hand. Although the water/cement
ratio of the screened concrete can be expected to remain unaltered, both the cement
content and water content increase, and generally an increase in strength has been
observed. For instance, screening out of particles greater than 19.05 mm ( in.) from
a mix with an original maximum size of 38.1 mm ( in.) has been found to increase
the compressive strength by 7 per cent, and the flexural strength by 15 per cent.12.45 On
another project, screening of the 38.1 to 152.4 mm ( to 6 in.) fraction has resulted in
an increase in compressive strength of 17 to 29 per cent.12.7 With air-entrained concrete,
wet screening produces some loss of air, and this causes an increase in strength.
These data reflect not only the effect of the change in the composition of the mix
but also the influence of the maximum size of aggregate per se (see p. 174).
The limited data on the effect of wet screening on the strength of concrete in direct
tension12.87 do not allow a generalized conclusion.
Test cores
The fundamental purpose of measuring the strength of concrete test specimens is to
estimate the strength of concrete in the actual structure. The emphasis is on the word
‘estimate’, and indeed it is not possible to obtain more than an indication of the
strength of concrete in a structure because this is dependent, inter alia, on the
adequacy of compaction and on curing. As shown earlier in this chapter, the strength
of a test specimen depends on its shape, proportions, and size, so that a test result does
not give the value of the intrinsic strength of the concrete. Nevertheless if, of two sets
of similar specimens made from two concretes, one set is stronger (at a statistically
significant level), it is reasonable to conclude that the concrete represented by this
specimen is stronger, too, There exist some methods of determining the strength of
concrete in situ, but the limitations on the interpretation of test results must be
remembered.
If the strength of standard compression test specimens is found to be below the
specified value, then either the concrete in the actual structure has too low a strength
as well, or else the specimens are not truly representative of the concrete in the
structure. This latter suggestion is often put forward in disputes on the acceptance, or
otherwise, of a suspect part of the structure: the test specimens may have been
disturbed while setting, they may have been exposed to frost before they hardened
sufficiently or have otherwise been improperly cured, or simply the results of the
compression test are doubted.
The argument is often resolved by testing a core of concrete taken from the suspect
member. If it is intended to determine the potential strength of the concrete mix used,
corrections for the actual conditions have to be applied. Cores can also be cut in order

663
to determine the actual strength of concrete in the structure. The distinction between
the two purposes must be clearly borne in mind when the test results are being
evaluated. The selection of the location of cores also depends on the purpose of
testing. This may be: to estimate the strength of a critical part of a structure, or of a
part suspected of having been damaged, for example, by frost; or alternatively, to
estimate a representative value for the entire structure, in which case a random
selection of locations is appropriate.
Cores can also be used to detect segregation of honeycombing or to check the bond
at construction joints or to verify the thickness of pavement.
Cores are cut by means of a rotary cutting tool with diamond bits. In this manner, a
cylindrical specimen is obtained, sometimes containing embedded fragments of
reinforcement, and usually with end surfaces far from plane and square. The core
should be soaked in water, capped, and tested in compression in a moist condition
according to BS EN 12504-1 : 2000 or ASTM C 42-04, but ACI 318-0212.124 specifies a
moisture condition corresponding to the service environment. Japanese
tests12.116 indicate that testing in a dry state yields strength values typically about 10 per
cent higher than when the cores are tested wet.
The influence of the height/diameter ratio of the cylinder on the recorded strength
was considered on p. 593. If the strength of cores is to be related to the strength of
standard cylinders (height/diameter ratio of 2) then, in the core, this ratio should be
near 2. When cubes are the standard test specimen, there is some advantage in using
cores with a height/diameter ratio of 1 because cylinders with this ratio have very
nearly the same strength as cubes. For values of the ratio between 1 and 2, a
correction factor has to be applied. Meininger et al.12.83 found the factor to be the same
for wet- and dry-tested cores, but lower than specified by ASTM C 42-04 (see Table
12.1).
Cores with height/diameter ratios lower than 1 give unreliable results, and BS EN
12504-1 : 2009 prescribes a minimum value of 0.95 prior to capping but, according to
BS 1881-120 : 1983, the cap thickness should not exceed 10 mm at any point. This
limitation should be observed although in practice, the length of the core may be
governed by the thickness of the concrete. Glueing cores which are too short is
possible.12.96
Use of small cores
Both British and ASTM Standards specify a minimum core diameter of 94 mm (3.7
in.) with the proviso that the core diameter be at least 3 times the maximum size of
aggregate; however, ASTM C 42-04 allows, as an absolute minimum, the ratio of the
two sizes to be 2.
Nevertheless, there exist circumstances where only very small cores can be drilled,
either because of the risk of structural damage or because of congestion of the
reinforcement or for aesthetic reasons. In such cases, some standards allow the use of
50 mm (2 in.) diameter cores. These small cores may violate the requirement of a
minimum ratio of core diameter to aggregate size, and the drilling operation can affect
the bond between the aggregate and the surrounding hardened cement
paste.12.98 Tests12.127 have shown that, when the maximum size of aggregate is 20 mm
( in.), 50 mm (2 in.) cores have a strength about 10 per cent lower than 100 mm (4
in.) cores; other tests12.110 on concretes with 28-day cube strengths between 20 and 60
MPa (or 3000 and 9000 psi) indicate that the difference is between 3 and 6 per cent. A

664
good correlation between the strength of 28 mm ( in.) diameter cores and the cube
strength was obtained in laboratory tests on concrete with a maximum size of
aggregate of 30 and 25 mm ( and 1 in.)12.78 (see Fig. 12.22).

Fig. 12.22. Relation between the strength of 28 by 28 mm ( by in.) cores and

the strength of 150 mm (6 in.) cubes; maximum aggregate size 25 and 30 mm12.78
Overall, in view of the numerous factors influencing the strength of cores, as
compared with the relative uniformity of cast standard compression test specimens,
the effect of core size can be considered to be unimportant. However, small cores
have a higher variability than standard-size cores; typical values12.100 of the coefficient
of variation are 7 to 10 per cent for 50 mm cores, and 3 to 6 per cent for 150 mm
cores. It follows that, for a given precision of the estimate of strength, the required
number of 50 mm cores is probably 3 times larger than the number judged adequate
for 100 mm (4 in.) or 150 mm (6 in.) cores. Likewise, when the core diameter is less
than three times the maximum size of aggregate, an increased number of cores has to
be tested.
Factors influencing strength of cores

665
The strength of cores is generally lower than that of standard cylinders, partly as a
consequence of the drilling operation and partly because site curing is
almost invariably inferior to curing prescribed for standard test specimens. However
careful the drilling, there is a high risk of slight damage. The effect appears to be
greater in stronger concrete, and Malhotra12.99 suggests that the reduction in strength
can be as high as 15 per cent for 40 MPa (6000 psi) concrete. A reduction of 5 to 7
per cent is considered reasonable by the Concrete Society.12.100
There is, however, a difficulty in separating out the effect of drilling because the
curing history of cores is perforce different from the curing history of cast test
specimens. The difficulty is exacerbated by the fact that the exact curing history of a
structure is usually difficult to determine so that the effect of curing on the strength of
cores is uncertain. For structures cured in accordance with the recommended practice,
Petersons12.67 found that the ratio of core strength to standard cylinder strength (at the
same age) is always less than 1, and decreases with an increase in the concrete
strength level. Approximate values of this ratio are: just under 1 when the cylinder
strength is 20 MPa (3000 psi) and 0.7 when it is 60 MPa (9000 psi).
Because cores are often taken after the 28-day test cylinders have been tested,
cylinders of an age comparable to the age of the cores may not be available, but it is
sometimes argued that cores taken from concrete many months old should have a
higher strength than at 28 days. This appears not to be the case in practice (see Figs
12.23 and 12.24), and there is evidence that in situ concrete often gains little in
strength after 28 days.12.102,12.103 Tests on high strength concrete12.112 show that, although
the strength of cores increases with age, the core strength, even up to the age of 1 year,
remains lower than the strength of standard 28-day cylinders; this is shown in Table
12.3.

666
Fig. 12.23. Development with time of strength of concrete cores made with Type I
cement expressed as a percentage of 28-day strength of standard cylinder (38
MPa (5500 psi)): (A) standard cylinder; (B) well-cured slab, core tested dry; (C)
well-cured slab, core tested wet; (D) poorly cured slab, core tested dry; (E)
poorly cured slab, core tested wet12.101

667
Fig. 12.24. Development with time of strength of concrete cores made with Type
III cement expressed as a percentage of 28-day strength of standard cylinder (38
MPa (5500 psi)): (A) standard cylinder; (B) well-cured slab, core tested dry; (C)
well-cured slab, core tested wet; (D) poorly cured slab, core tested dry; (E)
poorly cured slab, core tested wet12.101

Table 12.3. Development of the Strength of Cores* with Age (based on ref. 12.112)

*
Cores taken from columns cured using a sealing compound.

668
 These results accord with Petersons’ view12.104 that, for average conditions, the
increase in strength over that at 28 days is 10 per cent at three months, and 15 per cent
at the age of six months. The effect of age is, therefore, not easy to deal with but, in
the absence of definite moist curing, no increase in strength should be expected with
age and no age correction should be used in the interpretation of the strength of
cores.12.100
The location in the structure from which the core has been taken may affect the
strength of the core. If the core has been taken from concrete in tension, the core
strength may be low because of the presence of cracks:12.114 thus, a false picture of the
strength of the concrete in the structure can be obtained.
The position of the core with respect to the height of the lift may also be of
relevance. Cores usually have the lowest strength near the top surface of the structure,
be it a column, a wall, a beam, or even a slab. With an increase in depth below the top
surface, the strength of cores increases,12.67 but at depths greater than about 300 mm
there is no further increase. The difference can be as high as 10 or even 20 per cent. In
the case of slabs, poor curing increases this difference. Compressive and tensile
strengths are affected to the same degree.12.105 This pattern of strength is, however, not
universal, some tests indicating no significant variation in core strength with
height.12.112 It is likely that the variation in strength with height is the consequence of
trapped bleed water, coupled with a variation in compaction: when these factors are
absent, there is no variation in strength with height.
The presence of trapped bleed water may also be responsible, in part, for the
reported influence of the orientation of the core (vertical or horizontal) on its strength.
Cores drilled horizontally were found to have a strength lower by, typically, 8 per
cent.12.106 This effect is similar to the effect of bleed water on the strength of cubes (see
p. 590).
The conversion expressions of BS EN 12504-1 : 2009, distinguish between cores
drilled horizontally and those drilled vertically, the ratio of the strength of the former
to the latter being 0.92. However, if there is no trapped bleed water in the concrete,
the correction for horizontally-drilled cores may not be valid. It is also possible that
difficulties in horizontal drilling contribute to the lower strength of such cores.
British Standard BS EN 12504-1 : 2009 gives also correction factors which allow
for the weakening effect of transverse reinforcement in the core. Although some
effect of embedded steel on strength could be expected, the information on this is
contradictory. Reviews by Malhotra12.99 and by Loo et al.12.132 report some tests showing
no reduction in strength, and other tests where the reduction ranged between 8 and 18
per cent; the reduction seems to be higher when the height/diameter ratio of the core
is 2 than at lower values of this ratio.12.132 The Concrete Society12.100 also reports a
reduction in strength as a function of the position of the steel: the effect is greater the
further the steel is from the end of the core.
The tests of Loo et al.12.132 confirmed that embedded transverse reinforcement
reduces the strength of cores with a height/diameter ratio of 2, but the effect decreases
at low values of the height/diameter ratio; at a height/diameter ratio of 1, embedded
steel has no effect on the measured strength, regardless of the position of the steel in
the core. This effect is linked to the stress distribution in cylinders with various values
of the height/diameter ratio (see p. 595). When this ratio is 1, or in a cube, there is no
lateral tensile stress in the specimen, and the steel is well able to resist vertical
compression.

669
 In view of the various factors involved and of the conflicting data, no reliable
factor which allows for the presence of transverse steel can be accepted. The best
solution, if possible, is to take cores from a location such that they contain no
reinforcement, not only because it complicates the strength assessment, but more
importantly, because cutting reinforcement may have highly undesirable structural
consequences. In any case, the presence of steel parallel to the axis of the core is
unacceptable.
Relation of core strength to strength in situ
It should be emphasized that the core strengths, when converted to the strength of
cylinders of standard size or to cube strengths, represent, at best, the strength of in situ
concrete. They are not to be equated with the strength of standard test specimens,
which is the potential strength of the given concrete (see p. 584). Indeed, from the
preceding review of the various factors influencing the strength of cores, it is apparent
that it is not easy to interpret the strength of cores in relation to a specified 28-day
strength. Various reports12.99,12.103 suggest that, even under excellent conditions of placing
and curing, the strength of cores is unlikely to exceed 70 to 85 per cent of the strength
of standard test specimens. This view is supported by ACI 318-0812.124 which considers
that concrete in the part represented by a core test is adequate if the average strength
of 3 cores is equal to at least 85 per cent of the specified strength and if no single core
has a strength lower than 75 per cent of the specified value; no age allowance is made.
It should be noted that, according to ACI 318-95, cores are tested in a dry state if the
structure is dry in service, which should lead to a higher strength than when tested to
ASTM or British Standards (see p. 602). Thus, the requirements given above are
fairly liberal.
It is useful to note that the ‘85 per cent allowance’ is applied also to shotcrete
according to ACI 506.2-90.12.133 However, since shotcrete is accepted on the basis of
core strength, and not of moulded specimens, there is no logical reason for this
‘allowance’.12.111
In some cases, beam specimens can be sawn from road or airfield pavements,
using a diamond or silicon carbide saw. Such specimens are tested in flexure in
accordance with ASTM C 42-04 but, at least when siliceous aggregate is used, sawn
specimens give appreciably lower strengths than comparable moulded
beams.12.23 Cutting of beams is not much used and the means of obviating their use was
discussed on p. 600.
Cast-in-place cylinder test
It has been stressed repeatedly that standard compression test specimens give a
measure of the potential strength of concrete, and not of the strength of the concrete in
the structure. Knowledge of the latter cannot be directly obtained from tests on
separately made specimens. And yet, it is sometimes necessary to assess the strength
of concrete in the actual structure, for instance, for the purpose of deciding on the
time of removal of formwork, the application of prestress, or subjecting the structure
to loading. It may also be desired to assess the effectiveness of curing or of protection
from freezing.
One means of obtaining the requisite information is by use of cast-in-place
cylinder specimens which are made in push-out moulds. These special moulds are
fastened in tubular supports within the formwork of the structure prior to placing the
concrete, as shown in Fig. 12.25. This test method is limited to use in slabs with a

670
depth of 125 to 300 mm (5 to 12 in.) and is prescribed by ASTM C 873-04. The
mould is filled during placing of concrete in the slab formwork. Thus, the curing and
temperature conditions of the specimen and the slab are similar.12.122 Nevertheless, the
compaction of the concrete in the mould is not identical with the compaction of the
concrete in the actual structure. Consequently, the strength of cast-in-place cylinders
is reported by ASTM C 873-04 to be about 10 per cent higher than the strength of
cores drilled in the vicinity.

Fig. 12.25. Diagrammatic representation of a mould for a cast-in-place cylinder

The topic of the strength of concrete in structures is briefly considered on p. 625.
Influence of rate of application of load on strength
In the range of speeds at which a load can be applied to concrete, the rate of
application of load has a considerable effect on the apparent strength of concrete: the
lower the rate at which stress increases the lower the recorded strength. This is
probably caused by the increase in strain with time due to creep and, when limiting
strain is reached, failure takes place. Loading in compression over a period of 30 to
240 minutes has been found to cause failure at 84 to 88 per cent of the ultimate
strength obtained when the load is applied at the rate of approximately 0.2 MPa/s (30
psi/s).12.27 Concrete can withstand indefinitely only stresses up to about 70 per cent of
the strength determined under a load applied at the rate of 0.2 MPa/s (30 psi/s).12.28
Figure 12.26 shows that increasing the rate of application of compressive stress
from 0.7 kPa/s to 70 GPa/s (0.1 to 107 psi/s) doubles the apparent strength of concrete.
Raphael’s study12.52 of tests on concrete used in dams suggests that increasing the rate
of application of compressive stress by 3 orders of magnitude (which may be the case
in an earthquake) increases the strength by about 30 per cent. However, within the
practical range of rates of loading of compression specimens, that is between 0.07 and
0.7 MPa/s (10 and 100 psi/s), the measured strength varies only between 97 and 103
per cent of the strength at 0.2 MPa/s (30 psi/s).

671
 Fig. 12.26. Influence of the rate of application of load on the compressive
strength of concrete12.27
Nevertheless, for test results to be comparable, the stress has to be applied at a
standardized rate. The rate of loading of compression test specimens is prescribed by
ASTM C 39-09a as 0.20 to 0.30 MPa/s (28 to 42 psi/s), although a higher rate may be
applied during the first half of loading. British Standard BS EN 12390-3 : 2009
prescribes a rate of 0.2 to 1.0 MPa/s (30 to 145 psi/s) which has to be maintained
throughout the application of the load?
The results of flexure tests are affected by the speed of loading in a way similar to
compression tests. Increasing the rate of increase in stress in the extreme fibre of the
test beam from 2 to 130 kPa/s (0.3 to 19 psi/s) was found to increase the modulus of
rupture by about 15 per cent.12.20 The modulus of rupture increases linearly with the
logarithm of the rate of application of stress but, at very high rates of application of
tensile stress, there seems to be a departure from linearity: the rate of increase in
strength increases at an even greater rate. This is similar to the behaviour under
compressive stress (Fig. 12.26). At 170 MPa/s (24 700 psi/s) the modulus of rupture
was found to be 40 to 60 per cent larger than at 27 kPa/s (3.9 psi/s).12.27 British
Standard BS EN 12390-5 : 2009 prescribes a rate of increase in the extreme fibre in
flexure of between 0.04 and 0.06 MPa/s (5.8 to 8.8 psi/s) and ASTM C 78-09
specifies a rate between 0.86 and 1.21 MPa/min (2 to 3 psi/s).
It may be relevant to mention that the tensile strain capacity, which is of interest in
the control of cracking in mass concrete, depends on the rate of increase in tensile
stress. Liu and McDonald12.89 found that at slow rates of loading (0.17 MPa (25 psi) per
week) the strain capacity is 1.1 to 2.1 greater than when the rate of loading is 5 kPa/s
(0.68 psi/s). The magnitude of this increase, which is probably due to creep, depends
on the flexural strength and on the modulus of elasticity of the concrete: the increase
is greater for higher strengths and for lower values of the modulus of elasticity.12.89

672
 An increase in compressive strain capacity at lower rates of increase in strain was
reported by Dilger et al.12.68
The influence of the strain rate on the recorded strength is largest for direct tension,
intermediate for flexure, and least for compression12.54 (Fig. 12.27). Generally, stronger
concrete exhibits lower sensitivity to the strain rate.

Fig. 12.27. Influence of strain rate on relative strength (expressed as a proportion

of strength at the standard rate of strain) in tension, flexure, and compression
(based on ref. 12.54 with the permission of the publisher (ASCE))
Accelerated-curing test
Concrete is usually placed in a structure in stages or lifts, one on top of another. Thus,
by the time the results of the 28-day test, or even of the 7-day test, are available, a
considerable amount of concrete may overlay that represented by the test specimens
in question. It is then rather late for remedial measures if the concrete is too weak; if it
is too strong, this indicates that the mix used was uneconomical. Indeed, production
control with a 28-day delay is not sensible.
It is clear that it would be a tremendous advantage to be able to predict the 28-day
strength within a few hours of placing of concrete. The strength of concrete at 24
hours is an unreliable guide in this respect, not only because different blended
cements gain strength at varying rates, but also because even small variations
in temperature during the first few hours after casting have a considerable effect on
the early strength. It is, therefore, necessary for the concrete to have achieved a
greater proportion of its potential strength before testing, and a successful test based
on accelerated curing was developed by King12.46 in the mid-1950s. Since that time,
several accelerated-curing test methods have become standardized.
All these methods rely on accelerating the development of strength of standard
compression test specimens by a rise in temperature of the concrete specimen, without
permitting a loss of water from it. Details of the various tests are given in the
respective standards but a common feature of the tests is that, as in the conventional
strength tests, most of the test operations take place during usual working hours; this

673
is beneficial on construction projects where the site laboratory does not function
round the clock.
Four test methods using accelerated curing are prescribed by ASTM C 684-99 (03):
their brief description is given in Table 12.4. In Method A, the temperature rise is due
to the heat of hydration of cement, the primary function of the water bath being to
conserve that heat. In Method B, there is an additional input of heat from the boiling-
water bath. In Method C, curing takes place under adiabatic conditions, the sealed
specimen (so as to prevent moisture loss) being placed in an insulating container. In
Method D, a container pressurized to 10.3 MPa (1500 psi) at a temperature of 149 °C
(300 °F) is used. Thus, in Method D, specialized equipment is necessary;12.130 also, the
size of the test cylinder is limited so that, if the maximum size of aggregate is larger
than 25 mm (1 in.), wet sieving has to be resorted to.

Table 12.4. Summary of the Procedures for Accelerated Curing Prescribed by

ASTM C 684-99 (03)

A word of caution about the use of boiling water in Methods B and D should be
added: there is a danger of scalding and also of eye burns from a sudden escape of
steam.
There are three British methods, given in BS 1881-112 : 1983, all of which use a
water bath. One method is similar to Method A of ASTM C 684-99 (03), that is, it
uses a water bath at 35 °C. The second and third methods use water baths at 55 and
82 °C, respectively. In all cases, strength is determined at the age of up to 24 hours.
The British and American test methods differ with respect to the temperature of the
specimen when its strength is determined.
It is interesting to examine the effects of the specific curing procedures on the
products of hydration of cement. It is known that temperature influences the physical
characteristics of these products (see p. 361) but there is also a chemical effect in the
case of the boiling-water method: the crystallinity of ettringite is degraded.12.118 This
does not, however, affect the usefulness of the boiling-water method.

674
 The autogenous-curing method (Method C of ASTM C 684-99 (03) does not result
in a uniform acceleration of the development of strength because the nature of the
cement used controls the temperature rise, and this influences the rate of further
hydration. In addition, the strength is affected by the richness of the mix in a manner
different from that under normal curing. Nevertheless, a reliable relation between the
accelerated strength and the 28-day normal curing strength has been obtained. This is
of the form: 28-day strength = accelerated strength plus a constant.12.70
In fact, all the accelerated-curing test methods give a linear relation between the
accelerated strength and the strength of standard test specimens at 28 days, but each
method gives a different relation. Figure 12.28 shows an example of this relation for
Method B of ASTM C 684-99 (03), using a range of mixes containing fly ash from
different sources but only a single Portland cement.12.145 Generally speaking, the
specific equation relating the 28-day strength of standard specimens to the
accelerated-curing strength is different for cements having a different composition.
Some tests12.108 have shown that the maximum size of aggregate (but not its shape or
texture) also affects the relation.

675
 Fig. 12.28. Relation between the accelerated-curing strength according to
Method B of ASTM C 684-89 and the strength of standard test cylinders at 7 and
28 days12.145
According to BS 1881 : 112 : 1983, curing at 35 °C (95 °F) leads to a greater
sensitivity of the accelerated-curing strength to the variation in mix proportions. On
the other hand, tests on mortar indicate that curing at 35 °C (95 °F) has a high
reproducibility.12.118
To establish the relation between the accelerated-curing strengths and the 28-day
strength for the purpose of predicting the latter from the former, tests over a range of
strength values are necessary; ACI 214.1R-81 (Reapproved 1986)12.21 specifies the use
of at least three water/cement ratios. The correlation coefficient of such an equation is
generally very high so that the associated 95 per cent confidence interval is narrow: a
value of less than 3 MPa (400 psi) has been reported.12.120 This is so because the
accelerated-curing test is no more variable than the standard 28-day test.12.119
Accelerated-curing test methods can also be used for the determination of flexural
and splitting tension strengths.12.107
Direct use of accelerated-curing strength
The preceding reference to the variability of the results of the accelerated-curing tests
suggests a particularly worthwhile direct use of accelerated-strength testing in quality
control of the production of concrete: the early availability of test results makes it
possible fairly rapidly to adjust the mix proportions or make other changes in the
production process.
Furthermore, the fact that there is no unique relation between the accelerated-
curing strength and the standard 28-day strength begs the question as to whether the
purpose of determining the former strength should be to ‘predict’ the latter.
Admittedly, this was the original impetus for the development of accelerated-curing
test methods, but there is nothing sacrosanct about the 28-day strength, especially
when the specimens have been cured under ideal conditions, far removed from the
usual curing conditions of the concrete in situ. Moreover, the actual strength of the
concrete in the structure is influenced by the degree of compaction, bleeding, and
segregation. Thus, the 28-day strength of standard test specimens is no more
representative of the strength of the concrete in the actual structure than the strength
of specimens subjected to accelerated curing.
It is, therefore, strongly arguable that the accelerated-curing strength can, in its
own right, be used as an indication of the potential strength of the concrete which was
delivered for placement in the structure or, indeed, as a measure of the potential
strength. It is worth quoting Smith and Chojnacki12.69 who expressed the opinion that “a
suitable accelerated curing procedure can offer a more convenient and realistic way of
ascertaining if the concrete will satisfy the purpose for which it was designed”. This
was written in 1963, and the replacement of routine use of the standard 28-day
compression test by the accelerated-curing strength test is long overdue. The latter test
is superior as a quality control test, as well as a compliance test, because the outcome
of the test is available within a day or two of the placing of concrete.
The difficulty lies in the attachment of the engineers to the traditional test. For a
change to take place, the design ‘thinking’ would have to be entirely in accelerated-
strength values. As these are lower than the 28-day standard curing strength values,
there is a certain amount of reluctance to accept the new ‘numbers’. What should not

676
be done is to accept concrete in the first instance by the accelerated test and to require
it also to satisfy the 28-day cylinder test. This is too stringent a requirement because,
for a given variability of concrete, the probability of passing two tests is smaller than
that of passing either. Because the accelerated test and the 28-day test have
approximately the same variability, either of them alone is adequate to establish that
the concrete is from the desired population (see p. 639), and this is the aim of
acceptance testing.
Non-destructive tests
The tests described so far in this chapter involve specially made specimens which, as
such, do not necessarily give direct information about the concrete in the actual
structure; yet, this is what matters. Field-cured test specimens, and also cores, are of
some help in this respect. However, the former require pre-planning, and the latter
cause damage, albeit local, to the structure.
To get around these problems, a wide range of in-situ tests, known also as in-place
tests, have been developed. These tests are traditionally called non-destructive tests, it
being understood that some minor damage to the structure may be involved, although
its performance or appearance must not be impaired. An important feature of non-
destructive tests is that they permit re-testing at the same, or nearly the same, location
so that changes with time can be monitored.
The use of non-destructive tests leads to increased safety and allows better
scheduling of construction, thus making it possible to progress faster and more
economically. Broadly speaking, these tests can be categorized into those that assess
the strength of the concrete in situ, and those that determine other characteristics of
the concrete, such as voids, flaws, cracks, and deterioration.
With respect to strength, it should be noted that it can be only assessed, and not
measured, because the non-destructive tests are, for the most part, comparative in
nature. Thus, it is useful to establish an experimental relation between the property
being measured by a given test and the strength of test specimens or cores from the
actual concrete; thereafter, this relation can be used to ‘convert’ the non-destructive
test result into a strength value. An understanding of the physical relation between the
given non-destructive test result and strength is essential. This relation for the various
tests will be discussed in what follows. As this book is concerned with the properties
of concrete, and not with testing techniques, the actual details of the different tests
have to be sought in the relevant standards or handbooks.
One more general comment about the interpretation of the results of non-
destructive tests is necessary. The tests rarely give a ‘number’ which can be
unequivocally interpreted: engineering judgement is necessary. Thus, if the testing
arises from a dispute between the parties involved in the construction, the full test
programme should be determined in advance and the interpretation of possible test
results, bearing in mind their variability, should also be agreed. Otherwise, there is a
risk that one party or another will seek additional tests, and the dispute about the
concrete in the structure will be compounded by a dispute about testing. Helpful
advice about planning non-destructive testing is given in BS 1881-201 : 1986, and BS
6089 : 2010 gives a guide to the assessment of concrete strength in existing structures.
Rebound hammer test
This is one of the oldest non-destructive tests and it is still widely used. It was devised
in 1948 by Ernst Schmidt, and is therefore known also as Schmidt hammer,

677
or sclerometer, test. The hardness measured by the rebound hammer is quite different
from the hardness determined in tests on metals, which involve indentation.
The rebound hammer test is based on the principle that the rebound of an elastic
mass depends on the hardness of the surface against which the mass impinges.
However, despite its apparent simplicity, the rebound hammer test involves complex
problems of impact and the associated stress-wave propagation.12.134 In the rebound
hammer test (Fig. 12.29) a spring-loaded mass has a fixed amount of energy imparted
to it by extending the spring to a fixed position; this is achieved by pressing the
plunger against the surface of the concrete under test. Upon release, the mass
rebounds from the plunger, still in contact with the concrete surface, and the distance
travelled by the mass, expressed as a percentage of the initial extension of the spring,
is called the rebound number. This number is indicated by a rider moving along a
graduated scale. Some hammer models produce a print-out of test results. The
rebound number is an arbitrary measure because it depends on the energy stored in the
given spring and on the size of the mass. The hammer has to be used against a smooth
surface, preferably a formed one. Open-textured concrete cannot, therefore, be tested.
Trowelled surfaces should be rubbed smooth using a carborundum stone. If the
concrete under test does not form part of a larger mass, it has to be supported in an
unyielding manner, as jerking during the test would result in a lowering of the
rebound number recorded.

Fig. 12.29. Rebound hammer: (1) plunger; (2) concrete; (3) tubular housing; (4)
rider; (5) scale; (6) mass; (7) release button; (8) spring; (9) spring; (10) catch
The test is sensitive to local variations in the concrete; for instance, the presence of
a large piece of aggregate immediately underneath the plunger would result in an
abnormally high rebound number; conversely, the presence of a void in a similar
position would lead to a low result. Moreover, the energy absorbed by the concrete is
related both to its strength and its stiffness, so that it is the combination of strength
and stiffness that governs the rebound number.12.122 Because the stiffness of concrete is
influenced by the type of aggregate used (Fig. 12.30), the rebound number is not
uniquely related to the strength of concrete.

678
 Fig. 12.30. Relation between compressive strength and rebound number for
concrete cylinders made with different aggregates.12.48 (Readings taken on the side
of a cylinder with the hammer horizontal)
The plunger must always be normal to the surface of the concrete under test, but
the position of the hammer relative to the vertical will affect the rebound number.
This is due to the action of gravity on the travel of the mass in the hammer. Thus, the
rebound number of a floor is smaller than that of a soffit of the same concrete, and
inclined and vertical surfaces yield intermediate values. For this reason, and also
because of other factors which influence the rebound number, the use of ‘global’
diagrams relating the hardness number and strength is inadvisable. The correct
procedure is to establish experimentally the relation between the rebound number
measured on compression test specimens and their actual strength. If possible, the
specimen mould material should be the same as the formwork material in the structure.
While the position of the curves relating the compressive strength to the rebound
number varies, typically, for a change in strength of approximately 5 MPa (or 700 psi)
there is a change of 4 units in the rebound number. This relation is given only by way
of illustration and cannot be relied upon to detect small differences in strength. It
should be noted that different rebound hammers, even of the same design, cannot be
assumed necessarily to give the same rebound number.
In any case, the rebound hammer test measures the properties of only the surface
zone of concrete; according to BS 1881-202 : 1986 (withdrawn), the depth of this
zone is about 30 mm (or in.). Changes affecting only the surface of the concrete,
such as the degree of saturation at the surface (which lowers the rebound number,
see Fig. 12.3112.47) or carbonation (which raises the number12.125), have little influence on
the properties of the concrete at depth.

679
 Fig. 12.31. Relation between compressive strength of cylinders and rebound
number for readings with the hammer horizontal and vertical on a dry and a wet
surface of concrete12.47
Because of local variability in the hardness of concrete over a small area, the
rebound number should be determined at a number of locations in close proximity but,
according to ASTM C 805-08 not closer than 25 mm (1 in.) apart. British Standard
BS 1881-202 : 1986 (withdrawn) recommends testing on a grid pattern with a spacing
of 20 to 50 mm within an area not larger than 300 by 300 mm (12 by 12 in.); this
reduces the operator bias.
The rebound hammer test is largely comparative in nature and, as such, is useful in
the assessment of uniformity of concrete within a structure or in the manufacture of a
number of similar products such as precast elements. The test can also be used to
establish whether the rebound number has reached a value known to correspond to the
desired strength. This is of help in deciding when to remove falsework or to put the
structure into service. Another use of the hammer is to check whether the strength
development of a given concrete has been affected by frost at an early age but,
according to ASTM C 805-08, still frozen concrete may give very high rebound
numbers.
A particular application of the rebound hammer test is in assessing the abrasion
resistance of concrete floors, which largely depends on surface hardness.

680
 Overall, while the rebound hammer test is useful within a limited scope, the test is
not a strength test and exaggerated claims of its use as a replacement for the
compression test should not be accepted.
Penetration resistance test
The determination of the resistance of concrete to penetration by a steel rod, or probe,
driven by a fixed amount of energy can be used to assess the compressive strength of
concrete. The underlying principle is that, for standard test conditions, the depth of
penetration is inversely proportional to the compressive strength of concrete, but no
theoretical basis for this has been established. Moreover, the relation between strength
and the depth of penetration greatly depends on the hardness of the aggregate because
the coarse aggregate particles become fractured in the penetration tests, unlike in the
compression test. Specifically, softer aggregate allows greater penetration than hard
aggregate while the compressive strength may not be affected.12.122
The test equipment manufacturers supply ‘standard’ curves relating strength to the
depth of penetration for concretes containing coarse aggregates with various values of
hardness on Mohs’ scale. However, different investigators found significantly
different relations,12.126 possible contributing factors being the shape and surface
characteristics of the coarse aggregate.12.135 Thus, the relation between strength and
depth of penetration needs to be established by experiments for any given concrete.
There is, however, some difficulty even in this because the same cylinder or cube
cannot be used both for the penetration resistance test and for the compressive
strength test as the former test weakens the specimen. Moreover, if the penetration
resistance test is performed too close to the edge of the concrete, say less than 100 to
125 mm (4 to 5 in.), splitting can take place.
The test method for penetration resistance is prescribed by ASTM C 803-03 and by
BS 1881 : 207 : 1992. For convenience, it is not the actual depth of penetration, but
the complementary exposed length of a standard-length probe, that is measured. The
penetration probes are driven in sets of three, the average value being used as a test
result.
A typical relation between strength and the depth of penetration is shown in Fig.
12.32.

681
 Fig. 12.32. Influence of the hardness of aggregate on the relation between depth
of penetration and compressive strength (based on ref. 12.122)
The penetration resistance test is useful in determining whether formwork can be
removed. The test has some advantages over the rebound hammer test because a
greater depth of concrete is tested. Also, the number of tests required to detect, with
adequate confidence, a given difference in strength is reported12.140 to be smaller than
when the rebound hammer test is used. However, the cost of the penetration resistance
test is much higher. The penetration resistance test is likely to be preferable to drilling
small-diameter cores.
Pull-out test
This is a test which measures, by means of a special tension jack, the force required to
pull out a previously cast-in metal insert with an enlarged end (see Fig. 12.33). The
insert is pulled out with a lump of concrete, approximately in the shape of a frustum
of a cone. This shape is the consequence of the geometry of the insert together with
the bearing ring. For a given geometry, the pull-out force is related to the compressive
strength of concrete.

682
 Fig. 12.33. Diagrammatic representation of the pull-out test
This relation is purely experimental, and not based on consideration of stresses
involved, because the stress system at the fracture surface is three-dimensional: there
are radial and circumferential tensile stresses, and a compressive stress along the
surface of the cone.12.136 Consequently, the pull-out force should be reported as such (in
kN or lb), and calculations of a ‘pull-out strength’ lack reliable physical meaning. An
example of the relation between the pull-out force and the strength of cores for a wide
range of curing conditions is shown in Fig. 12.34.12.105

683
Fig. 12.34. Relation between compressive strength of cores and pull-out force for
actual structures12.105
The test method for the pull-out test is prescribed by ASTM C 900-06 and by BS
1881-207 : 1992. The ASTM Standard requires the depth of concrete above the
enlarged end of the insert to be equal to the diameter of the enlarged end; the Standard
also puts limits on the diameter of the bearing ring in relation to the diameter of the
enlarged end of the insert. These limits ensure that the apex angle of the frustum of
the cone lies between 54 and 70 degrees.12.122
According to Malhotra,12.113 the pull-out test is superior to the rebound hammer test
and to the penetration resistance test because a larger volume and a greater depth of
concrete are involved in the pull-out test. On the negative side, repair of concrete is
required. However, if the purpose of testing is to verify whether the concrete has
reached a desired strength, the pull-out test need not be carried out to completion: it
may be sufficient to apply a predetermined force to the embedded insert and, if it does
not pull out, the desired strength is assumed to exist.
Post-installed tests
A disadvantage of the pull-out test is that it requires pre-planned placing of inserts
prior to concreting. To make it possible to perform a pull-out test without prior

684
installation, several methods have been developed. These involve cutting a hole in the
hardened concrete, undercutting the hole with a special tool, and inserting an
expandable ring with a bolt attached to it. The pull-out test can then be performed in
the usual manner.12.139
Other post-installed tests include the internal fracture test, which has proved
useful in investigations of suspect concrete made with high-alumina cement.12.129 In this
test, a wedge-anchor bolt is tapped into a hole drilled into the concrete. The bolt is
then pulled by turning a nut resting on a thrust pad on a spherical seating. The torque
required to pull the bolt gives an assessment of the compressive strength of concrete,
although the bolt, when pulled, applies both vertical and horizontal forces to the
concrete.12.140 As in the case of the pull-out test, the pulling can be stopped at a
predetermined value of the torque, previously calibrated to correspond to a desired
strength. The internal facture test is described in BS 1881-207 : 1992.
In the break-off test, it is possible to assess the flexural strength of concrete in a
circular cross-section parallel to the surface of concrete. The section is formed by a
tube inserted into fresh concrete or by drilling a sleeve. A jack is used to apply a
transverse force in the element being broken off.12.138 The break-off test is standardized
in ASTM C 1150-90 (withdrawn) and BS 1881-207 : 1992.
There have also been developed pull-off tests involving measurement of the force
required to pull off a part of the concrete, using a glued metal disc.12.137 Thus, direct
tension is applied but the area upon which it acts is uncertain. The pull-off test is
prescribed in BS 1881 : 207 : 1992.
Tests involving removal of a fragment of concrete are proliferating. Good reviews
are offered by Bungey12.135 and by Carino.12.140
Ultrasonic pulse velocity test
This is a long-established, non-destructive test method which determines the velocity
of longitudinal (compressional) waves. This determination consists of measurement of
the time taken by a pulse – hence the name of the method – to travel a measured
distance. The apparatus includes transducers which are placed in contact with the
concrete, a pulse generator with a frequency of between 10 and 150 Hz, an amplifier,
a time measuring circuit, and a digital display of the time taken by the pulse of
longitudinal waves to travel between the transducers. The test method is prescribed by
ASTM C 597-09 and by BS EN 12504-4 : 2004.
The wave velocity, V, in a homogeneous, isotropic and elastic medium is related to
the dynamic modulus of elasticity, Ed, by the expression:

where ρ is density, and μ is Poisson’s ratio.

Concrete does not fulfil the physical requirements for the validity of the above
expression, and the determination of the modulus of elasticity of concrete from the
pulse velocity is not normally recommended.12.63 Nevertheless, Nilsen and
Aïtcin12.117 have found it useful to do so in monitoring the modulus of elasticity of high
strength concrete in service. It can be added that the value of Poisson’s ratio (see
p. 422) is generally not accurately known. However, a change in Poisson’s ratio over
the full range of possible values, that is from, say, 0.16 to 0.25, reduces the computed
value of the modulus by only about 11 per cent.

685
 With respect to the use of the value of the ultrasonic pulse velocity to determine
the strength of concrete, it has to be stated that there is no physical relation between
the two. It may be recalled that the modulus of elasticity is related to strength (see
p. 419), but this relation, too, has no physical basis. However, the ultrasonic wave
velocity is related to the density of concrete, as shown in the expression above. This
last-mentioned relation offers the rationale for the use of the ultrasonic wave velocity
measurements for the purpose of assessing the strength of concrete, but only subject
to strict limitations, discussed below.
The velocity of the ultrasonic pulse through concrete is the outcome of the time
taken by the pulse to travel through the hardened cement paste and through the
aggregate. The modulus of elasticity of aggregates varies considerably, so that the
pulse velocity of the concrete depends on the modulus of elasticity of the actual
aggregate and on the aggregate content of the mix. On the other hand, the strength of
concrete need not be significantly affected either by the content, or the modulus of
elasticity, of the aggregate. Consequently, no unique relation between ultrasonic pulse
velocity and compressive strength exists.12.62 Figure 12.35 shows that there is a
different relation for hardened cement paste, mortar, and concrete.

686
 Fig. 12.35. Relation between compressive strength and ultrasonic pulse velocity
for hardened cement paste, mortar, and concrete, in a dry and a moist condition
(based on ref. 12.62)
However, for a given aggregate and a given richness of the mix, the ultrasonic
pulse velocity of the concrete is affected by changes in the hardened cement paste,
such as a change in the water/cement ratio, which affects the modulus of elasticity of
the hardened cement paste. It is only within these limitations that the ultrasonic pulse
velocity test can be used to assess the strength of concrete. There is a further
limitation arising from the fact that the pulse travels faster through a water-filled void
than through an air-filled one. In consequence, the moisture condition of the concrete
affects the pulse velocity while the strength in situ is not affected (see Fig. 12.35).
Avoiding other spurious influences is also essential: for instance, steel
reinforcement, especially of large diameter, which lies along the path of the pulse
leads to an increased ultrasonic pulse velocity, but does not affect the compressive
strength of the concrete.12.135
Indeed, this is a particular case of the fundamental shortcoming of all non-
destructive tests in which the property of concrete being measured is affected by
various factors in a manner different from the influence of those factors on the
strength of concrete.
Despite the limitations listed above, the ultrasonic pulse velocity test has the
considerable merit of giving information about the interior of a concrete element. The
test is, therefore, useful to detect cracking (but not parallel to the direction of the
pulse), voids, deterioration due to frost or fire,12.61 and the uniformity of concrete in
similar elements. The test can be used for the purpose of following changes in a given
concrete element, for instance, due to repeated cycles of freezing and thawing. It is
useful to note that stress in concrete does not affect the value of ultrasonic pulse
velocity.12.142
The ultrasonic pulse velocity test can also be used to assess the strength of
concrete at very early ages, from about 3 hours onwards.12.146 This is of interest in
precasting or as an aid in deciding on removal of formwork, including steam-cured
concrete.12.143
An echo type of the ultrasonic pulse technique makes it possible to measure the
thickness of concrete roads and similar slabs.12.79
Further possibilities in non-destructive testing
We have discussed the various non-destructive test methods individually, but it is
possible to use more than one method at a time. This is advantageous when a variation
in properties of concrete affects the test results in opposite directions. Such is the case,
for instance, with the presence of moisture in concrete: an increase in the moisture
content increases the ultrasonic pulse velocity but decreases the rebound number
recorded by the rebound hammer.12.123 An example of the use of the combined results of
these two test methods is given in Fig. 12.36. Recommendations on the use of
combined non-destructive tests have been prepared by RILEM.12.141

687
 Fig. 12.36. Curves for the assessment of the in-situ compressive strength of
concrete using the combined ultrasonic pulse velocity and rebound hammer
tests12.123
There exist numerous other non-destructive tests on concrete in situ, some of
which are still in the developmental stage. These tests include radiography using
gamma rays or high-energy X-rays (to detect voids), radiometry (to measure
density), neutron transmission or reflection (to estimate the moisture content of
concrete), and surface-penetrating radar (to detect voids, cracks, or delamination). In
the impact–echo technique, transient stress waves induced by impact become
reflected by voids and cracks in concrete, and the resulting surface displacement is
monitored near the point of impact. Thus, flaws in the interior of concrete can be
detected.
The determination of acoustic emissions, which are transient elastic waves induced
by stresses representing a high proportion of the ultimate strength, can be used to
detect the development of cracking. The technique may be of value in assessing the
remaining integrity of a structure that has undergone extreme loading.12.66
The various tests referred to above are not discussed in this book as its scope is
limited to properties of concrete. However, one general comment should be made: all
test results are variable and should, therefore, be interpreted in the light of their
variability.
Resonant frequency method
In some cases, it is desirable to determine the progressive changes in the state of a
concrete specimen, for example, in consequence of repeated cycles of freezing and
thawing or of chemical attack. This can be done by determining the fundamental
resonant frequency of the specimen at appropriate stages of the investigation. From
this frequency, the dynamic modulus of elasticity of the concrete can be calculated.

688
 The vibration can be applied in a longitudinal, transverse (flexural) or torsional
mode. The test method is prescribed in ASTM C 215-08 and BS 1881-209 : 1990; the
latter standard covers the longitudinal mode only. In this mode, a specimen of
specified dimensions (preferably similar to those used to determine the modulus of
rupture) is clamped at its centre (Fig. 12.37) with a driving unit placed against one
end face of the specimen and a pick-up against the other. The exciter is driven by a
variable frequency oscillator with a range of 100 to 10 000 Hz. The vibrations
propagated within the specimen are received by the pick-up, amplified, and their
amplitude is measured by an appropriate indicator. The frequency of excitation is
varied until resonance is obtained at the fundamentel (i.e. lowest) frequency of the
specimen; this is indicated by the maximum deflection of the indicator.

Fig. 12.37. Test arrangement for the determination of the dynamic modulus of
elasticity in longitudinal vibration
If this frequency is n Hz, L is the length of the specimen, and ρ its density, then the
dynamic modulus of elasticity is given by:

Ed = Kn2L2ρ,

where K is a constant.
The length of the beam and its density have to be determined very accurately.
If L of a specimen of square cross-section is measured in millimetres and ρ in kg/m3,
then Ed in GPa is given by:

Ed = 4 × 10–15n2L2ρ.

If L is measured in inches, and ρ is in lb/ft3, then Ed in psi is:

Ed = 6 × 10–6n2L2ρ.

It should be emphasized that the dynamic modulus of elasticity calculated from the
resonance frequency cannot be interpreted to represent the strength of concrete; the
reasons for this were given in the section on the ultrasonic pulse velocity. It is only
under strictly limited circumstances of a single concrete mix that changes in strength
can be inferred from changes in the value of the modulus.
Tests on the composition of hardened concrete
In some disputes about the quality of hardened concrete, the question is raised
whether the composition of concrete was as specified and, to answer this, chemical
and physical tests are made on a sample of hardened concrete. The primary interest is
usually in the cement content and in the water/cement ratio, but the latter has to be
derived from determinations of the cement content and original water content.

689
 There are no universally applicable methods of chemical analysis because of the
wide range of materials used to make concrete. If the original mix ingredients are
available for testing, then the outcome of tests on a sample of hardened concrete is
fairly reliable but, even then, the interpretation of the results of analysis requires
engineering judgement based on practical experience.
Cement content
There exists no direct method of determining the content of cement, even Portland
cement alone, in a sample of concrete. The approach is to determine the contents of
soluble silica and calcium oxide, and hence to calculate the cement content. The lower
of the two values is used. The underlying basis is the fact that the silicates in Portland
cement are much more readily decomposed and made soluble than are the silica
compounds normally contained in aggregate. The same applies to the relative
solubilities of the lime compounds in the cement and in aggregate (excepting,
however, limestone aggregates), so that there exists also a soluble calcium oxide
method.
Standard methods for the determination of the content of Portland cement are
prescribed by ASTM C 1084-10 and BS 1881-124 : 1988, but the precision of the
results is generally too low to prove, or otherwise, compliance with the original
specification for the cement content; this is especially so in the case of mixes with low
cement contents, and it is often in this type of mix that the exact value of cement
content is required. Furthermore, the interpretation of the tests depends on the
knowledge of the chemical composition of aggregate. When large amounts of both
soluble silica and calcium oxide are liberated from the aggregate, the methods are
even less reliable.
A guide to tests when various cementitious materials are present is given in the
Concrete Society Report No. 32.12.25 That report suggests that it is possible to calculate
the slag content from a determination of the sulfide content in a sample of concrete,
provided the composition of the slag used is known, but reliable results are difficult to
obtain. No standard method for the determination of the fly ash content is available.
Likewise, the determination of the presence and dosage of admixtures is not routinely
possible because of the wide range of admixtures available and the low dosages
used.12.29
Determination of the original water/cement ratio
The water/cement ratio that existed at the time of placing of a concrete mix, now
hardened, can be calculated from the cement content (determined as described in the
preceding section) and from an estimate of the original water content. The original
water is the sum of the mass of combined water in the cement and of the volume of
capillary pores, which represents the remainder of the original water. The combined
water can be taken as being equal to 23 per cent of the mass of the cement (see p. 26)
or can be determined by igniting the sample at 1000 °C and measuring the water
driven off. The test method is prescribed by BS 1881: Part 124: 1989. According to
the Concrete Society Report No. 32,12.25 there is no evidence to show that this method
can be used for concretes made with blended cements. Even for Portland cement
concrete, the calculated water/cement ratio is likely to be within 0.1 of the actual
water/cement ratio.12.25 An estimate of this accuracy is of little practical value. Other
methods have been tried.12.147 The precision of the determination of the water/cement
ratio of hardened concrete is discussed in a recent paper by Neville.12.149

690
Physical methods
Guidance on petrographic examination of hardened concrete is given in ASTM C
856-04; ASTM C 457-10a covers other microscopical techniques which can be used
to determine the volumetric composition of a sample in the form of a polished slice.
These include the linear traverse method (see p. 558), the basis of which is the fact
that the relative volumes of the constituents of a heterogeneous solid are directly
proportional to their relative areas in a plane section, and also to intercepts of these
areas along a random line. The aggregate and the voids (containing air or evaporable
water) can be identified, the remainder being assumed to be hydrated cement. In order
to convert the quantity of the latter to the volume of unhydrated cement, we have to
know the specific gravity of dry cement and the non-evaporable water content of
hydrated cement (see p. 37). The test determines the cement content of the concrete
within 10 per cent, but the original water content or voids ratio cannot be estimated
since no distinction is made in the test between air and water voids.
The point-count method is based on the fact that the frequency with which a
constituent occurs at a given number of equally spaced points along a random line is a
direct measure of the relative volume of that constituent in the solid. Thus, a point
count by means of a stereomicroscope can rapidly give the volumetric proportions of
a hardened concrete specimen.
Variability of test results
The variation in strength of nominally similar test specimens has been mentioned, and
it follows that, whatever the test, the test results have to be interpreted in statistical
terms. The mere fact that some test result is, for example, larger than some other
result does not necessarily mean that the difference is significant and not a chance
consequence of the natural variability of values from the same source. While all test
results are variable, those derived from non-destructive tests generally have a larger
variability than is the case with standard compressive strength specimens. Some of the
simpler statistical terms will now be introduced.
Distribution of strength
Let us suppose that we have measured the compressive strength of 100 test specimens,
all made from similar concrete. This concrete can be imagined to be a collection of
units all of which could be tested; such a collection is referred to as the population,
and the portion of concrete in the actual test specimens is called the sample. It is the
purposes of the tests on the sample to supply information on the properties of the
parent population.
From the nature of the strength of concrete (p. 291) it would be expected that the
recorded strengths will be different for different specimens, i.e. the results will show a
scatter. To illustrate this let us consider the results on specimens tested in the
construction of an offshore platform,12.95 shown in Table 12.5. A good picture of the
distribution of these strengths can be obtained by grouping the actual strengths in
intervals of 1 MPa, so that we now have a certain number of specimens whose
strength falls within each interval, as in Table 12.5.

Table 12.5. Example of Distribution of Strength Test Results12.95

691
 If we now plot the (constant) strength interval as abscissae and the number of
specimens in each interval (known as frequency) as ordinates we obtain a histogram.
The area of the histogram represents the total number of specimens to an appropriate
scale. Sometimes, it is more convenient to express the frequency as a percentage of
the total number of specimens, i.e. to use relative frequency.
The histogram for the above-mentioned data is plotted in Fig. 12.38, and it can be
seen that it gives a clear picture of the scatter of results or, more accurately, of the
distribution of strength within the sample tested.

692
 Fig. 12.38. Histogram of strength values of p. 63712.95
Another simple measure of dispersion is given by the range of values, i.e. the
difference between the highest and the lowest strengths: 25 MPa in the above case.
The range is, of course, calculated extremely rapidly, but it is a rather crude measure:
it depends on two values only, and furthermore, in a large sample, these values are of
low frequency; thus range increases with sample size for the same underlying
distribution. The theoretical relation between range and standard deviation is shown
in Fig. 12.39, together with data obtained in practice.

693
 Fig. 12.39. Ratio of range to standard deviation for samples of different
size12.26 (Crown copyright)
If the number of specimens is increased indefinitely and, at the same time, the size
of the interval is decreased to a limiting value of zero, the histogram would become a
continuous curve, known as the distribution curve. For the strength of a certain type
of material, this curve would have a characteristic shape, and there are, in fact, several
‘type’ curves whose properties have been calculated in detail and are listed in
standard statistical tables.
One such type of distribution is the so-called normal or Gaussian distribution. The
applicability of this type of distribution to the strength of concrete was mentioned on
p. 605; the assumption of normal distribution is sufficiently close to reality to be an
extremely useful tool in computations (see Fig. 12.38).
The equation to the normal curve, which depends only on the values of the mean, μ,
and standard deviation, σ, is:

The standard deviation is defined in the next section. This equation is represented
graphically in Fig. 12.40, and it can be seen that the curve is symmetrical about the
mean value and extends to plus and minus infinity. This is sometimes mentioned as a
criticism of the use of normal distribution for strength, but the extremely low

694
probability of the occurrence of the very high or very low values is of little practical
significance.

Fig. 12.40. Normal distribution curve; percentage of specimens in intervals of

one standard deviation shown
The area under the curve between certain values of strength (measured in terms of
standard deviation) represents, in a manner similar to the histogram, the proportion of
specimens between the given limits of strength. Since, however, the curve refers to an
infinite population of specimens, and we deal with a limited number of them, the area
under the curve between given ordinates, expressed as a fraction of the total area
under the curve (and known therefore as proportional area), measures the chance that
the strength of an individual drawn at random, x, will lie between the given limits.
This chance multiplied by 100 gives the percentage of specimens that may be
expected in the long run to have a strength between the two limits considered.
Statistical tables give the values of proportional areas for different values of (x – μ)/σ.
Standard deviation
It can be seen from the foregoing discussion of probability that the dispersion of
strength about the mean is a fixed function of the standard deviation. This is defined
as the root-mean-square deviation, i.e.

where x represents the values of strength of all n specimens, and μ is the arithmetic
mean of these strengths, i.e. μ = ∑(x/n).
In practice, we deal with a limited number of specimens, and their mean, , is our
estimate of the true (population) mean μ. We calculate the deviations from and not
from μ, and therefore put (n – 1), instead of n, in the denominator of the expression
for the estimate of σ. The reason for this correction of n/(n – 1), known as Bessel’s

695
correction, is that the sum of squares of deviations has a minimum value when taken
about the sample mean, , and is therefore smaller than it would be if taken about the
population mean μ. (Bessel’s correction need not be applied when n is large.) Thus,
the estimate of σ is

An important practical point is that one value (e.g. the result of one specimen test)
yields no information whatever about the standard deviation and, therefore, about the
reliability or possible ‘error’ of the value obtained. Many calculators are programmed
to calculate the standard deviation direct but, for hand calculation, a more convenient
form of the expression for the standard deviation is

Thus, the sum of x2 is obtained without first finding the differences (x – ). Other
simplifications, such as subtracting a fixed quantity from all values, further aid
computation. To find s, Bessel’s correction is applied:

The standard deviation is expressed in the same units as the original variate, x, but
for many purposes it is convenient to express the scatter of results on a percentage
basis. We take then the ratio (σ/ ) × 100, and this is called the coefficient of variation.
It is a dimensionless quantity.
The graphical representation of the standard deviation (see Fig. 12.40) is the
horizontal distance from the mean to the point of contraflexure of the
normal distribution curve. Since the curve is symmetrical, the area under the curve
contained between abscissae μ – σ and μ + σ is 68 per cent of the total area under the
curve. In other words, the probability that the strength of a test specimen chosen at
random lies within the range μ ± σ is 0.68. The probabilities for other deviations from
the mean are indicated in Fig. 12.40.
For a given mean strength, the standard deviation characterizes fully the
distribution, assumed to be of the normal type; the variation in the value of the
standard deviation determines the spread of strengths in MPa or psi. It is useful to add
that the precision with which estimates the value of the population mean μ is
governed by the standard deviation of the mean, known as the standard error
σn where . Thus, there is a probability of 0.68 that is within the interval
μ ± σn .
The distribution curves for values of standard deviation of 2.5, 3.8, and 6.2 MPa
(350, 560, and 900 psi) are shown in Fig. 14.3. The value of the standard deviation
affects the (mean) strength that has to be aimed at in mix design for a given
‘minimum’ or characteristic strength specified by the designer of the concrete
structure. This problem is discussed fully in Chapter 14. Details of statistical methods
applicable to testing, particularly data on the choice of sample size, have to be sought

696
in specialized books.* The terms precision, repeatability and reproducibility are defind
in BS ISO 5725-1 : 1994.
*
E.g. J. B. KENNEDY and A. M. NEVILLE, Basic Statistical Methods for Engineers and
Scientists, 3rd Ed. (New York and London, Harper and Row, 1986).
References
12.1. CEB–FIP, Model Code 1990, 437 pp. (London, Thomas Telford, 1993).
12.2. R. C. MEININGER and N. R. NELSON, Concrete mixture evaluation and
acceptance for airfield pavements, in Airfield/Pavement Interaction: An
Integrated System, Proc. ASCE Conference, Kansas City, pp. 199–224
(ASCE, 1991).
12.3. A. M. NEVILLE, The influence of the direction of loading on the strength of
concrete test cubes, ASTM Bull. No. 239, pp. 63–5 (July 1959).
12.4. A. M. NEVILLE, The failure of concrete compression test specimens, Civil
Engineering, 52, No. 613, pp. 773–4 (London, July 1957).
12.5. H. F. GONNERMAN, Effect of end condition of cylinder on compressive strength
of concrete, Proc. ASTM, 24, Part II, p. 1036 (1924).
12.6. G. WERNER, The effect of type of capping material on the compressive
strength of concrete cylinders, Proc. ASTM, 58, pp. 1166–81 (1958).
12.7. U.S. BUREAU OF RECLAMATION, Concrete Manual, 8th Edn (Denver, Colorado,
1975).
12.8. R. L’HERMITE, Idées actuelles sur la technologie du béton, Documentation
Technique du Bâtiment et des Travaux Publics (Paris, 1955).
12.9. A. G. TARRANT, Frictional difficulty in concrete testing, The Engineer, 198,
No. 5159, pp. 801–2 (London, 1954).
12.10. A. G. TARRANT, Measurement of friction at very low speeds, The
Engineer, 198, No. 5143, pp. 262–3 (London, 1954).
12.11. P. J. F. WRIGHT, Compression testing machines for concrete, The
Engineer, 201, pp. 639–41 (London, 26 April 1957).
12.12. J. W. H. KING, Discussion on: Properties of concrete under complex states of
stress, in The Proc. Int. Conf. on the Structure of Concrete, p. 293 (London,
Cement and Concrete Assoc., 1968).
12.13. R. JONES, A method of studying the formation of cracks in a material
subjected to stress, British Journal of Applied Physics, 3, pp. 229–32
(London, 1952).
12.14. J. W. MURDOCK and C. E. KELSER, Effect of length to diameter ratio of
specimen on the apparent compressive strength of concrete, ASTM Bull., pp.
68–73 (April 1957).
12.15. K. NEWMAN, Concrete control tests as measures of the properties of
concrete, Proc. of a Symposium on Concrete Quality, pp. 120–38 (London,
Cement and Concrete Assoc., 1964).
12.16. R. H. EVANS, The plastic theories for the ultimate strength of reinforced
concrete beams, J. Inst. Civ. Engrs., 21, pp. 98–121 (London, 1943–44). See
also Discussion, 22, pp. 383–98 (London, 1943–44).

697
 12.17. U.S. BUREAU OF RECLAMATION 4914–92, Procedure for direct tensile strength,
static modulus of elasticity, and Poisson’s ratio of cylindrical concrete
specimens in tension, Concrete Manual, Part 2, 9th Edn, pp. 726–31
(Denver, Colorado, 1992).
12.18. A. M. NEVILLE, The influence of size of concrete test cubes on mean strength
and standard deviation, Mag. Concr. Res., 8, No. 23, pp. 101–10 (1956).
12.19. D. P. O’CLEARY and J. G. BYRNE, Testing concrete and mortar in
tension, Engineering, pp. 384–5 (London, 18 March 1960).
12.20. P. J. F. WRIGHT, The effect of the method of test on the flexural strength of
concrete, Mag. Concr. Res., 4, No. 11, pp. 67–76 (1952).
12.21. ACI 214.1R-81, Reapproved 1986, Use of accelerated strength testing, ACI
Manual of Concrete Practice, Part 2: Construction Practices and Inspection
Pavements, 4 pp. (Detroit, Michigan, 1994).
12.22. B. W. SHACKLOCK and P. W. KEENE, The comparison of compressive and
flexural strengths of concrete with and without entrained air. Cement Concr.
Assoc. Tech. Report TRA/283 (London, Dec. 1957).
12.23. S. WALKER and D. L. BLOEM, Studies of flexural strength of concrete – Part 3:
Effects of variations in testing procedures, Proc. ASTM, 57, pp. 1122–39
(1957).
12.24. P. J. F. WRIGHT, Comments on an indirect tensile test on concrete
cylinders, Mag. Concr. Res., 7, No. 20, pp. 87–96 (1955).
12.25. CONCRETE SOCIETY REPORT, Analysis of Hardened Concrete, Technical Report
No. 32, 111 pp. (London, 1989).
12.26. P. J. F. WRIGHT, Variations in the strength of Portland cement, Mag. Concr.
Res., 10, No. 30, pp. 123–32 (1958).
12.27. D. MCHENRY and J. J. SHIDELER, Review of data on effect of speed in
mechanical testing of concrete, ASTM Sp. Tech. Publ. No. 185, pp. 72–82
(1956).
12.28. W. H. PRICE, Factors influencing concrete strength, J. Amer. Concr. Inst., 47,
pp. 417–32 (Feb. 1951).
12.29. P. WITIER, Dosage des adjuvants dans les bétons durcis, Bulletin Liaison
Laboratoires Ponts et Chaussées, 158, pp. 45–52 (Nov.–Dec. 1988).
12.30. T. WATERS, The effect of allowing concrete to dry before it has fully
cured, Mag. Concr. Res., 7, No. 20. pp. 79–82 (1955).
12.31. S. WALKER and D. L. BLOEM, Effects of curing and moisture distribution on
measured strength of concrete, Proc. Highw. Res. Bd, 36, pp. 334–46 (1957).
12.32. R. H. MILLS, Strength–maturity relationship for concrete which is allowed to
dry, RILEM Int. Symp. on Concrete and Reinforced Concrete in Hot
Countries (Haifa, 1960).
12.33. W. S. BUTCHER, The effect of air drying before test: 28-day strength of
concrete, Constructional Review, pp. 31–2 (Sydney, Dec. 1958).
12.34. L. H. C. TIPPETT, On the extreme individuals and the range of samples taken
from a normal population, Biometrika, 17, pp. 364–87 (Cambridge and
London, 1925).

698
 12.35. A. M. NEVILLE, Some aspects of the strengths of concrete, Civil
Engineering (London), 54, Part 1, pp. 1153–5 (Oct. 1959); Part 2, pp. 1308–
11 (Nov. 1959); Part 3, pp. 1435–9 (Dec. 1959).
12.36. H. RÜSCH, Versuche zur Festigkeit der Biegedruckzone, Deutscher Ausschuss
für Stahlbeton, No. 120 (1955).
12.37. M. PRÔT, Essais statistiques sur mortiers et betons, Annales de l’Institut
Technique du Bâtiment et de Travaux Publics, No. 81, Béton, Béton Armé
No. 8, July–Aug. 1949.
12.38. R. F. BLANKS and C. C. MCNAMARA, Mass concrete tests in large cylinders, J.
Amer. Concr. Inst., 31, pp. 280–303 (Jan.–Feb. 1935).
12.39. W. J. SKINNER, Experiments on the compressive strength of anhydrite, The
Engineer, 207, Part 1, pp. 255–9 (13 Feb. 1959); Part 2, pp. 288–92 (London,
20 Feb. 1959).
12.40. H. F. GONNERMAN, Effect of size and shape of test specimen on compressive
strength of concrete, Proc. ASTM, 25, Part II. pp. 237–50 (1925).
12.41. A. M. NEVILLE, The use of 4-inch concrete compression test cubes, Civil
Engineering, 51, No. 605, pp. 1251–2 (London, Nov. 1956).
12.42. A. M. NEVILLE, Concrete compression test cubes, Civil Engineering, 52, No.
615, p. 1045 (London, Sept. 1957).
12.43. R. A. KEEN and J. DILLY, The precision of tests for compressive strength made
on -inch cubes of vibrated mortar, Cement Concr. Assoc. Tech. Report
TRA/314 (London, Feb. 1959).
12.44. B. W. SHACKLOCK, Comparison of gap- and continuously-graded concrete
mixes, Cement Concr. Assoc. Tech. Report TRA/240 (London, Sept. 1959).
12.45. S. WALKER, D. L. BLOEM and R. D. GAYNOR, Relationships of concrete strength
to maximum size of aggregate, Proc. Highw. Res. Bd, 38, pp. 367–79
(Washington DC, 1959).
12.46. J. W. H. KING, Further notes on the accelerated test for concrete, Chartered
Civil Engineer, pp. 15–19 (London, May 1957).
12.47. C. H. WILLETTS, Investigation of the Schmidt concrete test
hammer, Miscellaneous Paper No. 6-267 (U.S. Army Engineer Waterways
Experiment Station, Vicksburg, Miss., June 1958).
12.48. W. E. GRIEB, Use of the Swiss hammer for estimating the compressive
strength of hardened concrete, Public Roads, 30, No. 2, pp. 45–50
(Washington DC, June 1958).
12.49. K. SHIINA, Influence of temporary wetting at the time of test on compressive
strength and Young’s modulus of air-dry concrete, The Cement Association
of Japan Review, 36th General Meeting, pp. 113–5 (CAJ, Tokyo, 1982).
12.50. T. OKAJIMA, T. TSHIKAWA and K. ICHISE, Moisture effect on the mechanical
properties of cement mortar, Transactions of the Japan Concrete Institute, 2,
pp. 125–32 (1980).
12.51. S. POPOVICS, Effect of curing method and final moisture condition on
compressive strength of concrete, ACI Journal, 83, No. 4, pp. 650–7 (1986).
12.52. J. M. RAPHAEL, Tensile strength of concrete, Concrete International, 81, No.
2, pp. 158–65 (1984).

699
 12.53. W. T. HESTER, Field testing high-strength concretes: a critical review of the
state-of-the-art, Concrete International, 2, No. 12, pp. 27–38 (1980).
12.54. W. SUARIS and S. P. SHAH, Properties of concrete subjected to impact, Journal
of Structural Engineering, 109, No. 7, pp. 1727–41 (1983).
12.55. D. N. RICHARDSON, Review of variables that influence measured concrete
compressive strength, Journal of Materials in Civil Engineering, 3, No. 2,
pp. 95–112 (1991).
12.56. H. KUPFER, H. K. HILSDORF and H. RÜSCH, Behavior of concrete under biaxial
stresses. J. Amer. Concr. Inst., 66, pp. 656–66 (Aug. 1969).
12.57. K. NEWMAN and L. LACHANCE, The testing of brittle materials under uniform
uniaxial compressive stress, Proc. ASTM, 64, pp. 1044–67 (1964).
12.58. H. HANSEN, A. KIELLAND, K. E. C. NIELSEN and S. THAULOW, Compressive
strength of concrete – cube or cylinder? RILEM Bull. No. 17, pp. 23–30
(Paris, Dec. 1962).
12.59. B. L. RADKEVICH, Shrinkage and creep of expanded clay–concrete units in
compression, CSIRO Translation No. 5910 from Beton i Zhelezobeton, No. 8,
pp. 364–9 (1961).
12.60. Z. PIATEK, Wlansości wytrzymalościowe i reologiczne keramzytobetonu
konstrukcyjnego, Arch. Inz. Ladowej, 16, No. 4, pp. 711–29 (Warsaw, 1970).
12.61. H. W. CHUNG and K. S. LAW, Diagnosing in situ concrete by ultrasonic pulse
technique, Concrete International, 5, No. 10, pp. 42–9 (1983).
12.62. V. R. STURRUP, F. J. VECCHIO and H. CARATIN, Pulse velocity as a measure of
concrete compressive strength, in In Situ/Nondestructive Testing of Concrete,
Ed. V. M. Malhotra, ACI SP-82, pp. 201–27 (Detroit, Michigan, 1984).
12.63. R. E. PHILLEO, Comparison of results of three methods for determining
Young’s modulus of elasticity of concrete, J. Amer. Concr. Inst., 51, pp.
461–9 (Jan. 1955).
12.64. V. M. MALHOTRA, Effect of specimen size on tensile strength of concrete, J.
Amer. Concr. Inst., 67, pp. 467–9 (June 1970).
12.65. A. M. NEVILLE, A general relation for strengths of concrete specimens of
different shapes and sizes, J. Amer. Concr. Inst., 63, pp. 1095–109 (Oct.
1966).
12.66. P. F. MLAKER et al., Acoustic emission behavior of concrete, in In
Situ/Nondestructive Testing of Concrete, Ed. V. M. Malhotra, ACI SP-82, pp.
619–37 (Detroit, Michigan, 1984).
12.67. N. PETERSONS, Should standard cube test specimens be replaced by test
specimens taken from structures?, Materials and Structures, 1, No. 5, pp.
425–35 (Paris, Sept.–Oct. 1968).
12.68. W. H. DILGER, R. KOCH and R. KOWALCZYK, Ductility of plain and confined
concrete under different strain rates, ACI Journal, 81, No. 1, pp. 73–81
(1984).
12.69. P. SMITH and B. CHOJNACKI, Accelerated strength testing of concrete
cylinders, Proc. ASTM, 63, pp. 1079–101 (1963).
12.70. P. SMITH and H. TIEDE, Earlier determination of concrete strength
potential, Report No. RR124 (Department of Highways, Ontario, Jan. 1967).

700
 12.71. C. BOULAY and F. DE LARRARD, A new capping system for testing HPC
cylinders: the sand-box, Concrete International, 15, No. 4, pp. 63–6 (1993).
12.72. P. M. CARRASQUILLO and R. L. CARRASQUILLO, Evaluation of the use of current
concrete practice in the production of high-strength concrete, ACI Materials
Journal, 85, No. 1, pp. 49–54 (1988).
12.73. D. N. RICHARDSON, Effects of testing variables on the comparison of neoprene
pad and sulfur mortar-capped concrete test cylinders, ACI Materials
Journal, 87, No. 5, pp. 489–502 (1990).
12.74. P. M. CARRASQUILLO and R. L. CARRASQUILLO, Effect of using unbonded capping
systems on the compressive strength of concrete cylinders, ACI Materials
Journal, 85, No. 3, pp. 141–7 (1988).
12.75. AUSTRALIAN PRE-MIXED CONCRETE ASSN, An Investigation into Restrained
Rubber Capping Systems for Compressive Strength Testing of Concrete,
Technical Bulletin 92/1, 59 pp. (Sydney, Australia, 1992).
12.76. E. C. HIGGINSON, G. B. WALLACE and E. L. ORE, Effect of maximum size of
aggregate on compressive strength of mass concrete, Symp. on Mass
Concrete, ACI SP-6, pp. 219–56 (Detroit, Michigan, 1963).
12.77. U.S. BUREAU OF RECLAMATION, Effect of maximum size of aggregate upon
compressive strength of concrete, Laboratory Report No. C-1052 (Denver,
Colorado, June 3, 1963).
12.78. F. INDELICATO, A statistical method for the assessment of concrete strength
through micropores, Materials and Structures, 26, No. 159, pp. 261–7
(1993).
12.79. H. MAILER, Pavement thickness measurement using ultrasonic
techniques, Highway Research Record, 378, pp. 20–8 (1972).
12.80. P. ROSSI and X. WU, Comportement en compression du béton: mécanismes
physiques et modélisation, Bulletin Liaison Laboratoires Ponts et
Chaussées, 189, pp. 89–94 (Jan.–Feb. 1994).
12.81. U.S. ARMY CORPS OF ENGINEERS, Standard CRD-C 62-69: Method of testing
cylindrical test specimens for planeness and parallelism of ends and
perpendicularity of sides, Handbook for Concrete and Cement, 6 pp.
(Vicksburg, Miss., 1 Dec. 1969).
12.82. K. L. SAUCIER, Effect of method of preparation of ends of concrete cylinders
for testing, U.S. Army Engineers Waterways Experiment Station Misc. Paper
No. C-7-12, 19 pp. (Vicksburg, Miss. April 1972).
12.83. R. C. MEININGER, F. T. WAGNER and K. W. HALL, Concrete core strength – the
effect of length to diameter ratio. J. Testing and Evaluation, 5, No. 3, pp.
147–53 (May 1977).
12.84. J. G. WIEBENGA, Influence of grinding or capping of concrete specimens on
compressive strength test results, TNO Rep. No. BI-76-71/01.571.104,
Netherlands Organization for Applied Scientific Research, 5 pp. (Delft, 26
July 1976).
12.85. G. SCHICKERT, On the influence of different load application techniques on the
lateral strain and fracture of concrete specimens, Cement and Concrete
Research, 3, No. 4, pp. 487–94 (1973).

701
 12.86. D. J. HANNANT, K. J. BUCKLEY and J. CROFT, The effect of aggregate size on the
use of the cylinder splitting test as a measure of tensile strength, Materials
and Structures, 6, No. 31, pp. 15–21 (1973).
12.87. NIANXIANG XIE and WENYAN LIU, Determining tensile properties of mass
concrete by direct tensile test, ACI Materials Journal, 86, No. 3, pp. 214–19
(1989).
12.88. K. W. NASSER and A. A. AL-MANASEER, It’s time for a change from 6 × 12- to
3 × 6-in. cylinders, ACI Materials Journal, 84, No. 3, pp. 213–16 (1987).
12.89. T. C. LIU and J. E. MCDONALD, Prediction of tensile strain capacity of mass
concrete, J. Amer. Concr. Inst., 75, No. 5, pp. 192–7 (1978).
12.90. R. L. DAY and N. M. HAQUE, Correlation between strength of small and
standard concrete cylinders, ACI Materials Journal, 90, No. 5, pp. 452–62
(1993).
12.91. V. KADLEČEK and Z. ŠPETLA, Effect of size and shape of test specimens on the
direct tensile strength of concrete, RILEM Bull., No. 36, pp. 175–84 (Paris,
Sept. 1967).
12.92. R. J. TORRENT, A general relation between tensile strength and specimen
geometry for concrete-like materials, Materials and Structures, 10, No. 58,
pp. 187–96 (1977).
12.93. A. BAJZA On the factors influencing the strength of cement compacts, Cement
and Concrete Research, 2, No. 1, pp. 67–78 (1972).
12.94. Z. P. BAžANT et al., Size effect in Brazilian split-cylinder tests: measurements
and fracture analysis, ACI Materials Journal, 88, No. 3, pp. 325–32 (1989).
12.95. J. MOKSNES, Concrete in offshore structures, Concrete Structures –
Norwegian Inst. Technology Symp., Trondheim, Oct. 1978, pp. 163–76
(1978).
12.96. U. BELLANDER, Concrete strength in finished structures; Part 1, Destructive
testing methods. Reasonable requirements, CBI Research 13:76, 205 pp.
(Swedish Cement and Concrete Research Inst., 1976).
12.97. P. ROSSI et al., Effet d’échelle sur le comportement du béton en
traction, Bulletin Liaison Laboratoires des Fonts et Chaussées, 182, pp. 11–
20 (Nov.–Dec. 1992).
12.98. J. H. BUNGEY, Determining concrete strength by using small-diameter
cores, Mag. Concr. Res., 31, 107, pp. 91–8 (1979).
12.99. V. M. MALHOTRA, Contract strength requirements – cores versus in
situ evaluation. J. Amer. Concr. Inst., 74, No. 4, pp. 163–72 (1977).
12.100. CONCRETE SOCIETY, Concrete core testing for strength, Technical Report No.
11, 44 pp. (London, 1976).
12.101. R. D. GAYNOR, One look at concrete compressive strength, NRMCA Publ. No.
147, National Ready Mixed Concrete Assoc, 11 pp. (Silver Spring,
Maryland, Nov. 1974).
12.102. J. M. PLOWMAN, W. F. SMITH and T. SHERRIFF, Cores, cubes and the specified
strength of concrete, The Structural Engineer, 52, No. 11, pp. 421–6 (1974).

702
 12.103. W. E. MURPHY, Discussion on paper by V. M. Malhotra: Contract strength
requirements – core versus in situ evaluation, J. Amer. Concr. Inst., 74, No.
10, pp. 523–5 (1977).
12.104. N. PETERSONS, Recommendations for estimation of quality of concrete in
finished structures, Materials and Structures, 4, No. 24, pp. 379–97 (1971).
12.105. U. BELLANDER, Strength in concrete structures, CBI Reports 1:78, 15 pp.
(Swedish Cement and Concrete Research Inst., 1978).
12.106. J. R. GRAHAM, Concrete performance in Yellowtail Dam,
Montana, Laboratory Report No. C-1321 U.S. Bureau of Reclamation,
(Denver, Colorado, 1969).
12.107. V. M. MALHOTRA, An accelerated method of estimating the 28-day splitting
tensile and flexural strengths of concrete, Accelerated Strength Testing, ACI
SP-56, pp. 147–67 (Detroit, Michigan, 1978).
12.108. R. S. AL-RAWI and K. AL-MURSHIDI, Effects of maximum size and surface
texture of aggregate in accelerated testing of concrete, Cement and Concrete
Research, 8, No. 2, pp. 201–9 (1978).
12.109. J. W. GALLOWAY, H. M. HARDING and K. D. RAITHBY, Effects of Moisture
Changes on Flexural and Fatigue Strength of Concrete, Transport and Road
Research Laboratory, No. 864, 18 pp. (Crowthorne, U.K., 1979).
12.110. W. E. YIP and C. T. TAM, Concrete strength evaluation through the use of
small diameter cores, Mag. Concr. Res., 40, No. 143, pp. 99–105 (1988).
12.111. S. GEBLER and R. SCHUTZ, IS 0.85 valid for shotcrete?, Concrete
International, 12, No. 9, pp. 67–9 (1990).
12.112. R. L. YUAN et al., Evaluation of core strength in high-strength
concrete, Concrete International, 13, No. 5, pp. 30–4 (1991).
12.113. V. M. MALHOTRA, Evaluation of the pull-out test to determine strength of in-
situ concrete, Materials and Structures, 8, No. 43, pp. 19–31 (1975).
12.114. A. SZYPULA and J. S. GROSSMAN, Cylinder vs. core strength, Concrete
International, 12, No. 2, pp. 55–61 (1990).
12.115. W. C. GREER, JR., Variation of laboratory concrete flexural strength
tests, Cement, Concrete and Aggregates, 5, No. 2, pp. 111–22 (Winter,
1983).
12.116. S. YAMANE, et al. Concrete in finished structures, Takenaka Tech. Res. Rept.
No. 22, pp. 67–73 (Tokyo, Oct. 1979).
12.117. A. U. NILSEN and P.-C. AÏTCIN, Static modulus of elasticity of high-strength
concrete from pulse velocity tests, Cement, Concrete and Aggregate, 14, No.
1, pp. 64–6 (1992).
12.118. K. MATHER, Effects of accelerated curing procedures on nature and properties
of cement and cement-fly ash pastes, in Properties of Concrete at Early
Ages, ACI SP-95, pp. 155–71 (Detroit, Michigan, 1986).
12.119. J. F. LAMOND, Quality assurance using accelerated strength testing, Concrete
International, 5, No. 3, pp. 47–51 (1983).
12.120. J. ÖZETKIN, Accelerated strength testing of Portland–pozzolan cement
concretes by the warm water method, ACI Materials Journal, 84, No. 1, pp.
51–4 (1987).

703
 12.121. F. M. BARTLETT and J. G. MACGREGOR, Effect of moisture condition on
concrete core strengths, ACI Materials Journal, 91, No. 3, pp. 227–36
(1994).
12.122. ACI 228.1R-89, In-place methods for determination of strength of
concrete, ACI Manual of Concrete Practice, Part 2: Construction Practices
and Inspection Pavements, 25 pp. (Detroit, Michigan, 1994).
12.123. U. BELLANDER, Concrete strength in finished structures; Part 3, Non-
destructive testing methods. Investigations in laboratory and in-situ, CBI
Research 3:77, p. 226 (Swedish Cement and Concrete Research Inst., 1977).
12.124. ACI 318-02, Building code requirements for structural concrete, ACI
Manual of Concrete Practice, Part 3: Use of Concrete in Buildings – Design,
Specifications, and Related Topics, 443 pp.
12.125. S. AMASAKI, Estimation of strength of concrete in structures by rebound
hammer, CAJ Proceedings of Cement and Concrete, No. 45, pp. 345–51
(1991).
12.126. R. S. JENKINS, Nondestructive testing – an evalution tool, Concrete
International, 7, No. 2, pp. 22–6 (1985).
12.127. C. JAEGERMANN and A. BENTUR, Development of destructive and non-
destructive testing methods for quality control of hardened concrete on
building sites and in precast factories, Research Report No. 017-196, Israel
Institute of Technology Building Research Station (Haifa, July 1977).
12.128. K. M. ALEXANDER, Comments on “an unsolved mystery in concrete
technology”, Concrete, 14, No. 4. pp. 28–9 (London, April 1980).
12.129. A. J. CHABOWSKI and D. W. BRYDEN-SMITH, Assessing the strength of in-
situ Portland cement concrete by internal fracture tests, Mag. Concr. Res., 32,
No. 112, pp. 164–72 (1980).
12.130. K. W. NASSER and R. J. BEATON, The K-5 accelerated strength tester, J. Amer.
Concr. Inst., 77, No. 3, pp. 179–88 (1980).
12.131. L. M. MELIS, A. H. MEYER and D. W. FOWLER, An Evaluation of Tensile
Strength Testing, Research Report 432-1F, Center for Transportation
Research, University of Texas, 81 pp. (Austin, Texas, Nov. 1985).
12.132. Y. H. Loo, C. W. TAN and C. T. TAM, Effects of embedded reinforcement on
measured strength of concrete cylinders, Mag. Concr. Res., 41, No. 146, pp.
11–18 (1989).
12.133. ACI 506.2-90, Specification for materials, proportioning, and application of
shotcrete, ACI Manual of Concrete Practice, Part 5: Masonry, Precast
Concrete, Special Processes, 8 pp. (Detroit, Michigan, 1994).
12.134. T. AKASHI and S. AMASAKI, Study of the stress waves in the plunger of a
rebound hammer at the time of impact, in In Situ/Nondestructive Testing of
Concrete, Ed. V. M. Malhotra, ACI SP-82, pp. 19–34 (Detroit, Michigan,
1984).
12.135. J. H. BUNGEY, The Testing of Concrete in Structures, 2nd Edn, 222 pp.
(Surrey University Press, 1989).
12.136. W. C. STONE and N. J. CARINO, Comparison of analytical with experimental
internal strain distribution for the pullout test, ACI Journal, 81, No. 1, pp. 3–
12 (1984).

704
 12.137. J. H. BUNGEY and R. MADANDOUST, Factors influencing pull-off tests on
concrete, Mag. Concr. Res., 44, No. 158, pp. 21–30 (1992).
12.138. M. G. BARKER and J. A. RAMIREZ, Determination of concrete strengths with
break-off tester, ACI Materials Journal, 85, No. 4, pp. 221–8 (1988).
12.139. C. G. PETERSEN, LOK-test and CAPO-test development and their
applications, Proc. Inst. Civ. Engrs, Part 1, 76, pp. 539–49 (May 1984).
12.140. N. J. CARINO, Nondestructive testing of concrete: history and challenges,
in Concrete Technology: Past, Present, and Future, V. Mohan Malhotra
Symposium, ACI SP-144, pp. 623–80 (Detroit, Michigan, 1994).
12.141. RILEM Committee 43, Draft recommendation for in-situ concrete strength
determination by combined non-destructive methods, Materials and
Structures, 26, No. 155, pp. 43–9 (1993).
12.142. S. POPOVICS and J. S. POPOVICS, Effect of stresses on the ultrasonic pulse
velocity in concrete, Materials and Structures, 24, No. 139, pp. 15–23
(1991).
12.143. G. V. TEODORU, Mechanical strength property of concrete at early ages as
reflected by Schmidt rebound number, ultrasonic pulse velocity, and
ultrasonic attenuation, in Properties of Concrete at Early Ages, ACI SP-95,
pp. 139–53 (Detroit, Michigan, 1986).
12.144. S. NILSSON, The tensile strength of concrete determined by splitting tests on
cubes, RILEM Bull. No. 11, pp. 63–7 (Paris, June 1961).
12.145. K. W. NASSER and V. M. MALHOTRA, Accelerated testing of concrete:
evaluation of the K-5 method, ACI Materials Journal, 87, No. 6, pp. 588–93
(1990).
12.146. R. H. ELVERY and L. A. M. IBRAHIM, Ultrasonic assessment of concrete
strength at early ages, Mag. Concr. Res., 28, No. 97, pp. 181–90 (Dec. 1976).
12.147. B. MAYFIELD, The quantitative evaluation of the water/cement ratio using
fluorescence microscopy, Mag. Concr. Res., 42, No. 150, pp. 45–9 (1990).
12.148. E. ARIOGLU and O. S. KOYLUOGLU, Discussion of ‘Are current concrete strength
tests suitable for high strength concrete?’, Materials and Structures, 29, No.
193, pp. 578–80 (1996).
12.149. A. M. NEVILLE, HOW closely can we determine the water-cement ratio of
hardened concrete? Materials and Structures, 36, pp. 311–18 (June 2003).
12.150. BEATTIE, A., Lightweight aggregates: benefits and practicalities, The
Structural Engineer, 88, pp. 14–18 (Dec. 2010).

705
Chapter 13. Concretes with particular properties

This chapter deals with several types of concretes which can be used when particular
properties are required. The term ‘particular’ does not imply anything unusual or
rarely required; rather, it refers to specific properties which are desirable under the
given circumstances. Several types of concretes will be considered. First, concretes
containing the different cementitious materials which are frequently used nowadays
(discussed in Chapter 2). These are: fly ash, ground granulated blastfurnace slag, and
silica fume.13.90
The second type of concrete to be considered is the so-called high performance
concrete. This concrete invariably contains one or more of the cementitious materials
mentioned above, and usually a superplasticizer as well. The term ‘high performance’
is somewhat pretentious because the essential feature of this concrete is that its
ingredients and proportions are specifically chosen so as to have particularly
appropriate properties for the expected use of the structure; these properties are
usually a high strength or a low permeability.
The third, and last, type of concrete discussed in this chapter is lightweight
concrete, that is, concrete with a density significantly lower than the density of
concretes made with normal aggregates, which is in the range of 2200 to 2600
kg/m3 (140 to 160 lb/ft3).
One more type of concrete should be mentioned: high-density concrete, which is
used for the purpose of attenuation of high-energy X-rays, gamma rays, and neutrons.
Because of this specialized use of high-density concrete, it will not be considered in
the present book.
Concretes with different cementitious materials
The preceding chapters have dealt with concretes which may contain a range of
cementitious materials, but mainly with concretes containing Portland cement only.
The reason for this approach is that, until fairly recently, Portland cement was
considered as the ‘best’, if not the sole, cementitious material in concrete. When other
materials, primarily fly ash and ground granulated blastfurnace slag, were introduced,
they were viewed as replacements or substitutes for cement, and their influence and
performance were judged against the standard of concrete containing only Portland
cement.
The situation has changed dramatically: as pointed out on p. 90, several
cementitious materials are today concrete ingredients in their own right. These
materials, fly ash, ground granulated blastfurnace slag (for brevity, referred to
as ggbs), and silica fume, were discussed in Chapter 2 in so far as their physical and
chemical properties are concerned. When various properties of concrete were
considered in subsequent chapters, the influence of these materials was often
mentioned. This, however, was unavoidably fragmentary, and it is now proposed to
review the properties of concretes containing the various cementitious materials.
It can be argued that the influences of the individual cementitious materials should
be discussed first. On the other hand, a brief review of these materials considered
together is useful in painting a general picture of their role in the behaviour of
concrete. Therefore, the common features of two, or all three, of these materials and

706
the use of more than one of them at a time will be discussed. This will be followed by
their individual consideration.
General features of use of fly ash, ggbs, and silica fume
An argument which is sometimes advanced in favour of the use of these various
cementitious materials is that, compared with Portland cement, they save energy and
conserve resources. This is factually correct, but it is the actual technical benefits of
the inclusion of these materials in the concrete that are the strongest argument in
favour of their use. Indeed, in many cases, they should be used in preference to a
Portland-cement-only mix regardless of economic or environmental considerations.
There is some difficulty in presenting the available information on the influence
and use of the three cementitious materials – fly ash, ggbs, silica fume – in an
objective and generally valid way. An extremely large number of research papers
have been published but, in many of them, an enthusiastic researcher describes a
single set of tests on one of these materials and points out the benefits of the use of
that particular material, which often is a specific local product. This description may
well be a true and factual account, but the conclusions are usually couched in terms of
a comparison with a ‘reference’ mix containing Portland cement only. The differences
between the mix with the given cementitious material and the ‘reference’ mix may
include workability, strength at some age or another, total content of cementitious
material or water/cement ratio; any one of these may be of importance in construction.
A worthwhile generalization from such a comparison is not possible. What is of use is
a general review of the pattern of properties of the mixes containing the different
cementitious materials. This should make it possible to assess the properties of
concretes with different ingredients, possibly in different proportions. Specific
properties of any given mix have to be ascertained by experiment.
The various cementitious materials affect the progress of hydration in consequence
of their chemical composition, reactivity, particle size distribution, and particle
shape.13.9 The actual reactivity of ggbs depends on its composition, glass content, and
particle size.13.9 High-calcium fly ash (ASTM Class C, BS EN class W) is much more
reactive than Class F (BS EN Class V) fly ash and, therefore, exhibits some similarity
to the behaviour of ggbs.13.9 The reaction of Class F fly ash requires a high alkalinity of
the pore water. This alkalinity is reduced when silica fume or ggbs are present in the
mix. In consequence, the reactivity of fly ash in such mixes is reduced.13.15
At a given total content of cementitious material the inclusion of fly ash or ggbs
generally reduces the water demand and improves workability. In the case of ggbs,
the improvement may not be measurable in terms of slump but, once vibration has
started, concrete containing ggbs becomes ‘mobile’ and compacts well. Silica fume
greatly reduces, or even eliminates, bleeding. The improvement of workability by fly
ash is ascribed to the spherical shape of its particles. However, the inclusion of fly ash
and, to a lesser extent, of ggbs in the mix has the physical effect of modifying the
flocculation of cement, with a resulting reduction in the water demand.13.9 The changed
dispersion of cement particles is reflected in the microstructure of the hydrated
cement paste, mainly its pore size distribution, the median pore size being smaller and,
consequently, the permeability being lower.13.9 This effect is present at a constant total
porosity (which is controlled by the overall water/cement ratio).
The improvement of the strength of concrete by fly ash is not only the
consequence of its pozzolanicity but also of the ability of the very small fly ash
particles to ‘fit in’ between cement particles. Proof of this is provided by the

707
beneficial effect of fly ash used with Portland blastfurnace cement when pozzolanic
reaction is unlikely13.12
Durability aspects
Although an early reason for the use of the various cementitious materials in concrete
was their influence on the rate of development of heat and of strength, even more
important is their influence on the resistance of concrete to chemical attack, which is
the consequence not only of the chemical nature of the hydrated cement paste but also
of its microstructure. This topic was considered in Chapters 10 and 11. It is no
exaggeration to say that the cementitious materials have a major influence on all
aspects of durability related to the transport of attacking agents through concrete. A
reason for this is that, generally, the cementitious materials considered in this chapter
are finer than Portland cement and, therefore, improve particle packing, so that,
provided adequate wet curing is applied, their presence reduces permeability.13.92
Even though the use of fly ash or ggbs reduces permeability, it allows faster
carbonation.13.113 The increase in the rate of carbonation is greater when fly ash is used
with Portland blastfurnace cement.13.12 When the ggbs plus fly ash content is more than
60 per cent, the increase in carbonation is greater the greater the fly ash
content.13.13 The enhanced carbonation need not necessarily be large in practice when
mixes with proper mix proportions are used. Also, carbonation may reduce the
permeability, but not when both fly ash and ggbs are present in the mix.13.12 Good
resistance to freezing and thawing without air entrainment was found in concretes
(with a water/cement ratio of 0.27 and a superplasticizer) containing Class C fly ash
representing 20 to 35 per cent of the mass of the total cementitious material, and silica
fume (10 per cent on the same basis).13.11 Likewise, good resistance to sulfate attack
was observed with Class C fly ash contents up to 50 per cent and 10 per cent of silica
fume.13.11
Control of the alkali–silica reaction is a specialized topic in which a detailed
knowledge of the aggregate to be used is necessary (see p. 144). However, the
beneficial effects of the incorporation of fly ash (about 30 to 40 per cent by mass) or
of ggbs (about 40 to 50 per cent by mass) in the blended cement should be
noted.13.7 These materials contain only a small amount of water-soluble alkalis so that,
at a given content of cementitious material which includes Portland cement with a
high alkali content, the presence of ggbs or fly ash in the blended cement reduces the
total alkali content in the mix.13.10 Thus, the use of these materials may obviate the need
for low-alkali cement but the absence of expansive reactions should be verified by
tests.
The beneficial effects of the inclusion of silica fume in steam-cured concrete at
65 °C (149 °F) upon its penetrability by chlorides was confirmed by Campbell and
Detwiler.13.4 For significant improvement, the minimum silica fume content was 10 per
cent in Portland-cement-only concrete, but 7.5 per cent was highly effective in mixes
containing 30 to 40 per cent of ggbs in the total cementitious material.13.4 It may be
added that curing Portland-cement-only concrete at 50 °C (122 °F) was found to result
in increased penetrability by chlorides.13.3
Further studies by Detwiler et al.13.2 confirmed the beneficial effect of inclusion of
both silica fume and ggbs in concrete cured at 50 and 70 °C (122 and 168 °F) upon
penetrability of chlorides. These findings were obtained on concretes with
water/cement ratios of 0.40 and 0.50 and with silica fume and ggbs contents of 5 and
30 per cent, respectively, by mass of the total cementitious material. Generalizations

708
of optimum contents or proportions are not possible because the penetrability of the
resulting concrete is affected by the degree of hydration at the time of exposure to
chlorides. Information on the influence of inclusion of both silica fume and fly ash in
concrete cured at high temperatures upon its permeability to chlorides does not seem
to be available.
Variability of materials
The three cementitious materials discussed in this chapter are not manufactured
specifically for use in concrete but are industrial by-products. This situation is
reflected in their variability.
Fly ash is a by-product of burning pulverized coal to generate electric power. The
power station operators are aware of the commercial value of a uniform fly ash, but
periodic variations in the operation of a power station (especially if not a base supply
station) can result in occasionally varying properties of the fly ash. There are, of
course, also differences in the fly ash produced by different power stations. Moreover,
even the same power station will produce fly ash with varying properties if the coal
used is non-uniform in the short- or long-term. Ash classification and beneficiation
would be helpful but they would increase the cost of the fly ash.
It follows that the users of fly ash have to be aware of the properties of the actual
material used in concrete, and they cannot rely on standardized assumptions about the
particle size distribution of fly ash or its carbon content. In consequence, a simple
picture of the behaviour of concrete containing fly ash cannot be presented because
fly ash is not a single material of nearly constant composition. Fly ashes are rather
like the various Types of Portland cements in that the fly ashes also have a range of
physical and chemical characteristics. It is, therefore, not surprising that the use of fly
ash, especially because its content in concrete can vary widely, results in a range of
effects.
On the other hand, slag, being a by-product of a highly controlled process (see
p. 79), is much less variable; the same applies to silica fume.
Returning to the subject of fly ash, we should note that the hydration of any given
fly ash depends on the chemical properties and the fineness of the Portland cement in
the mix. It is not surprising that there is no simple relation between the proportion of
fly ash in the total cementitious material and the properties of the resulting concrete of
otherwise fixed proportions. Inevitably, attempts to relate, by a single equation, the
strength of concrete, even of fixed proportions, to the various properties of fly ash
such as fineness, residue of particles above a certain size, pozzolanic indices, carbon
content, glass content, and chemical composition, have been unsuccessful.13.6 Indeed,
this situation is to be expected, given that no single equation can predict the strength
properties of Portland cements alone from their physical and chemical properties.
Fly ash and ggbs are very valuable ingredients of concrete. They are also
economically advantageous because they are by-products of other processes, and are
continuously available – indeed, in need of being disposed of. It is worth reflecting
that, in consequence of changes in our industrial patterns, especially in the
consumption of iron and in the sources of energy, less fly ash and less slag may
become available in the future (see also p. 656). New cementitious materials might
need to be developed.
Concrete containing fly ash

709
A brief description of the physical and chemical properties of fly ash was presented
in Chapter 2. We shall now consider the use of fly ash in concrete and discuss the
properties of the resulting concrete; a further discussion of the properties of the fly ash
itself, in so far as they affect the properties of concrete, will also be included.
The importance of fly ash cannot be exaggerated: it is no longer a cheap substitute
for cement, nor an ‘extender’ or an addition to the mix. Fly ash bestows important
advantages upon concrete, and it is, therefore, essential to understand the role and
influence of fly ash.
The variability of the properties of fly ash was mentioned in the preceding section.
This variability arises from the fact that fly ash is not a specially manufactured
product and cannot, therefore, be governed by strict requirements of a standard. The
main influences are the nature of the coal and the manner of its pulverization, the
operation of the furnace, the process of precipitation of ash from the combustion gases,
and especially the extent of classification of the particles in the exhaust system. Even
when all these are constant, a power station which varies its operation in response to
the power demand produces a variable fly ash; this is not so with a base-load power
station. The variations in the fly ash are those in glass content, carbon content, particle
shape and size distribution, as well as in the presence of magnesia and other minerals,
and even in colour. It is possible to improve the size distribution of fly ash particles
by classification and by grinding.
As just mentioned, the burning process of pulverized coal influences the shape of
the fly ash particles. High temperature favours the formation of spherical particles, but
the need to reduce the emission of NOx gases requires the use of lower peak burning
temperatures so that the minerals with a high melting point do not always fuse
completely. A consequence of this is a reduction in the proportion of spherical
particles of fly ash and also in the proportion of particles smaller than 10 μm; however,
the proportion of particles larger than 45 μm is not affected.13.12,13.34 These changes
militate against the beneficial effects of fly ash in concrete. Thus, there is need for
changes in technology which will satisfy both the NOx emission requirements and the
particle properties desirable from the standpoint of their use in concrete.
It should be pointed out, however, that, in most countries, much uniform and
excellent fly ash for use in concrete is consistently produced, and there is no doubt
that, world-wide, the consumption of fly ash in concrete increases and is expected to
continue to do so. What is not possible is to provide information about a ‘standard’, or
even typical, fly ash. Consequently, specific guidance on the use of fly ash as a
generic material cannot be presented.
Influence of fly ash on properties of fresh concrete
The main influence is that on water demand and on workability. For a constant
workability, the reduction in the water demand of concrete due to fly ash is usually
between 5 and 15 per cent by comparison with a Portland-cement-only mix having the
same cementitious material content; the reduction is larger at higher water/cement
ratios.13.12
A concrete mix containing fly ash is cohesive and has a reduced bleeding capacity.
The mix can be suitable for pumping and for slipforming; finishing operations of fly
ash concrete are made easier.
The influence of fly ash on the properties of fresh concrete is linked to the shape of
the fly ash particles. Most of these are spherical and solid, but some of the large

710
particles are hollow spheres, known as cenospheres, or are vesicular and irregular in
shape.
The reduction in water demand of concrete caused by the presence of fly ash is
usually ascribed to their spherical shape, this being called a ‘ball-bearing effect’.
However, other mechanisms are also involved and may well be dominant. In
particular, in consequence of electrical charges, the finer fly ash particles become
adsorbed on the surface of the cement particles. If enough fine fly ash particles are
present to cover the surface of the cement particles, which thus become deflocculated,
the water demand for a given workability is reduced.13.156 An amount of fly ash in
excess of that required to cover the surface of the cement particles would confer no
further benefit with respect to water demand. Indeed, the reduction in water demand
becomes larger with an increase in the fly ash content only up to about 20 per
cent.13.156 The effect of fly ash is not additional to the action of superplasticizers. Thus,
it seems likely that the action of fly ash, like that of superplasticizers, on water
demand is through dispersion and adsorption of the fly ash on the particles of Portland
cement.13.156 Malhotra recommends the use of fly ash above 50 per cent of cementitious
material but this is not universally accepted.13.160
The presence of carbon in fly ash was referred to on p. 85. One consequence of a
high carbon content in fly ash is that it adversely affects workability. Variation in
carbon content may also lead to erratic behaviour with respect to air entrainment,
some air-entraining agents becoming adsorbed by the porous carbon particles.
Fly ash in the mix has a retarding effect, typically of about 1 hour, probably caused
by the release of present at the surface of the fly ash particles. The retardation
may be advantageous when concreting in hot weather; otherwise, an accelerator may
be needed. Only initial setting is delayed, the time interval between setting and final
stiffening being unaffected.
The delayed setting is additional to the retarding effect of some admixtures by
lower temperatures. This may lead to blistering and delamination.13.160 The preceding
statement is not an argument against using a high content of fly ash in the
cementitious material, but simply as an indication of the need to establish the
properties of such concrete in the presence of admixtures.
Hydration of fly ash
Pozzolanic reactions were considered in Chapter 2. In the case of fly ash, the products
of reaction closely resemble C-S-H produced by hydration of Portland cement.
However, the reaction does not start until sometime after mixing. In the case of Class
F fly ash (see p. 85), this can be as long as one week or even more. An explanation of
this delay, offered by Fraay et al.13.15 is as follows. The glass material in fly ash is
broken down only when the pH value of the pore water is at least about 13.2, and the
increase in the alkalinity of the pore water requires that a certain amount of hydration
of the Portland cement in the mix has taken place. Moreover, the reaction products of
Portland cement precipitate on the surface of the fly ash particles, which act as nuclei.
When the pH of the pore water becomes high enough, the products of reaction of
the fly ash are formed on the fly ash particles and in their vicinity. A consequence of
these early reactions is that their products often remain in the shape of the original
spheres of fly ash. With the passage of time, further products diffuse away and
precipitate within the capillary pore system; this results in a reduction in the capillary
porosity and, consequently, a finer pore structure (see Fig. 13.1).13.15

711
Fig. 13.1. Change in pore size distribution (determined by mercury porosimetry)
in cement paste containing 30 per cent of Class F fly ash by mass of total
cementitious material (based on ref. 13.15)
The sensitivity of the fly ash reaction to the alkalinity of the pore water means that
the reactivity of fly ash is influenced by the alkali content of the Portland cement with
which the fly ash is to be used. (This is, however, disproved by Osbæck.13.114) For
example, because rapid-hardening Portland (Type III) cement leads to a more rapid
development of alkalinity of pore water than ordinary Portland cement, the pozzolanic
reaction of fly ash starts earlier when Type III cement is used. The preceding
observations illustrate the complexity of the behaviour of fly ashes which makes
generalizations difficult and points to the need for tests involving both the fly ash and
the Portland cement which are to be used together.
A consequence of the delay in the reactions of fly ash is the beneficial pattern of
heat evolution by hydration (see Chapter 8).
Further progress of the pozzolanic reaction of Class F fly ash is slow: the presence
of as much as 50 per cent of unreacted fly ash after one year is quoted by Fraay et
al.13.15
Whereas Portland-cement-only concrete with a medium or a high water/cement
ratio, under suitable storage conditions, continues to gain strength over a long period,
this is not so when fly ash is incorporated in the mix. No further strength development
beyond the age of 3 to 5 years was found in concretes with water/cement ratios of 0.5
to 0.8; the Class F fly ash content, expressed as a percentage of the mass of the total
cementitious material, ranged from 47 to 67.13.16,13.17
Class C fly ash (BSEN Class W) (see p. 85) which has a high lime content, reacts,
to some extent, direct with water; in particular, some C2S may be present in the fly

712
ash13.157 and this compound reacts to form C-S-H. Also, crystalline C3A and other
aluminates are reactive.13.9 In addition, as with Class F fly ash, there is a reaction of
silica with calcium hydroxide produced by the hydration of Portland cement. Thus,
Class C fly ash reacts earlier than Class F fly ash, but some Class C fly ashes do not
show a long-term increase in strength.13.18
Because the reactions of fly ash in concrete take a long time, prolonged wet curing
is essential. A consequence of this is that tests on compression specimens cured under
standard wet conditions may be misleading with respect to the strength of concrete in
situ. This, of course, is also the case with Portland-cement-only concrete, but the
influence of curing on strength is more pronounced when fly ash is included in the
mix.
Higher temperature, between 20 and 80 °C (68 and 176 °F) accelerates the
reactions of fly ash to a greater extent than is the case with Portland cement alone.
However, the usual retrogression of strength follows (cf. p. 361).13.21 The reduction in
strength with an increase in temperature between 200 and 800 °C is also similar to, or
possibly even greater than, that in concrete made with Portland cement only.13.20
Because the reactivity of fly ash sharply increases with an increase in temperature,
the behaviour of concrete containing fly ash may be different in massive sections
(where hydration of the Portland cement component raises the temperature) from the
behaviour in small concrete elements at room temperature.13.9 This observation is
relevant to any prediction of the rate of gain of strength of concrete containing fly ash.
Strength development of fly ash concrete
The test method of ASTM C 311-07 provides for the measurement of strength of
mortars containing fly ash representing 20 per cent by mass of the total cementitious
material and establishes a strength activity index. However, as already discussed, the
reactions of fly ash are affected by the properties of Portland cement with which it is
used. Moreover, in addition to the effect of chemical reactions, fly ash has a physical
effect of improving the microstructure of the hydrated cement paste. The main
physical action is that of packing of the fly ash particles at the interface of coarse
aggregate particles, which are absent in the mortar used in the test of ASTM C 311-
07.13.12
For these reasons, strength activity measurements do not adequately establish the
contribution of fly ash to the development of strength of a particular concrete in
which the fly ash is to be incorporated. This is an example of the inappropriateness of
tests on mortar for the purpose of establishing the effect of a given factor on concrete.
The extent of packing depends both on the fly ash and on the cement used: better
packing is achieved with coarser Portland cement and with finer fly ash.13.12 One
beneficial effect of packing on strength is a reduction in the volume of entrapped air
in the concrete,13.12 but the main contribution of packing lies in a reduction in the
volume of large capillary pores.
It is worth noting that the positive influence of the fineness of fly ash is coupled
with its spherical shape. Therefore, grinding of fly ash, although it increases fineness,
may result in the destruction of spherical particles, with a consequent increase in
water demand of the mix due to the irregular angular shape of the fly ash particles.13.26
Control of particle size of fly ash is usually effected on the basis of residue larger
than 45 μm (No. 325 ASTM) sieve, but this is not sufficiently discriminatory with

713
respect to the reactivity of fly ash and its contribution to strength development in
concrete.
Typically, about one-half of the particles in fly ash are smaller than 10 μm, but
there may be wide variations. It is particles of that size that are most reactive.13.22 The
reactivity is very high when the median diameter of fly ash particles is smaller still: 5
or even 2.5 μm.
As far as the coarse particles of fly ash are concerned, Idorn and
Thaulow13.23 suggested that these particles can be considered as ‘microaggregate’
which improves the density of the hydrated cement paste in a manner similar to the
effect of unhydrated remnants of Portland cement particles. This is beneficial with
respect to strength, resistance to crack propagation, and stiffness. The resulting system
of capillary pores is better able to retain water which can be available for long-term
hydration.13.23
The glass content of the fly ash strongly affects its reactivity. In the case of Class C
fly ash, the lime content is also a factor influencing reactivity. However, knowledge
of these characteristics does not make it possible to predict the performance of any
given fly ash, and tests are necessary; tests with the actual Portland cement to be used
are preferable.
It was mentioned on p. 656 that the beneficial influence of fly ash upon water
demand does not extend beyond a fly ash content of 20 per cent by mass. An
excessive content of fly ash is not beneficial from the point of view of strength
development either. The limiting content is probably around 30 per cent by mass of
total cementitious material, as can be seen from Fig. 13.2.13.19

714
Fig. 13.2. Influence of content of fly ash in the cementitious material (by mass) on
strength of hardened cement paste13.19
As has been repeatedly stated, quantified predictions of the influence of fly ash on
strength are not possible. For example, the data of Fig. 13.2 can be contrasted with the
apparent lack of a positive influence of fly ash upon strength even as late as one year,
which was reported by the Portland Cement Association.13.14
Average values of strength of concrete cylinders moist cured at 23 °C (73 °F)
(obtained from tests on six Class F fly ashes and four Class C fly ashes) are shown
in Table 13.1.13.14 All the mixes had a total cementitious material content of 307
kg/m3 (517 lb/yd3) with a 25 per cent content of fly ash by mass of total cementitious
material. The water/cement ratio was 0.40 to 0.45, and the mixes had a slump of 75
mm. The same table gives the strength of a Portland-cement-only concrete with the
same cement content and the same water/cement ratio. It is worth adding that the

maximum size of aggregate was 9.5 mm ( in.) so that the beneficial effect of fly ash
with respect to packing around the coarse aggregate particles was smaller than would

715
be the case with conventional concrete; therein may lie the explanation of the
apparently limited effect of fly ash on strength.

Table 13.1. Typical Compressive Strength of Fly Ash Concretes13.14

In this connection, it should be noted that, because the specific gravity of fly ash is
much lower than that of Portland cement (typically 2.35 as compared with 3.15), for
the same mass, the volume of fly ash is about 30 per cent higher than that of cement.
This must be taken into account in determining the mix proportions of concrete:
usually, a lower content of fine aggregate is used than with Portland-cement-only
concrete.
As for physical properties of concrete other than strength, it appears that creep and
shrinkage are not fundamentally affected by the use of fly ash.
Durability of fly ash concrete
As discussed in Chapters 10 and 11, the selection of ingredients of a concrete mix
must include consideration of their effect on durability. As in the case of strength,
much depends on the actual fly ash used.
One consequence of the slow reaction of fly ash in the concrete is that, initially, the
concrete has a higher permeability than concrete with a similar water/cement ratio (on
the basis of the total cementitious material) but containing Portland cement only.
However, with time, fly ash concrete acquires a very low permeability.13.15 It is,
nevertheless, essential that the concrete containing fly ash undergoes prolonged
curing. The detrimental effect of inadequate curing on the absorption properties of the
outer zone of concrete is greater the higher the fly ash content.13.101 This effect is even
more pronounced than the effect on the strength of concrete containing fly ash. Thus,
reliance on strength alone may not be adequate for the purpose of assessing the
durability of fly ash concrete in cases where penetration of concrete by aggressive
agents is critical.
With respect to the resistance to sulfate attack, it should be noted that alumina and
lime in the fly ash may contribute to the sulfate reactions. Specifically, when present
in the glass part of the fly ash, alumina and lime provide a long-term source of

716
material which can react with sulfates to form expansive ettringite.13.25 A high
silica/alumina ratio probably reduces the vulnerability to sulfate attack13.28 but no
reliable generalization is possible.
It seems that inclusion of Class F fly ash in concrete improves its sulfate resistance,
probably mainly through the removal of calcium hydroxide. The content of fly ash
should generally be between 25 and 40 per cent of the total cementitious material.
Reliable information on the behaviour of Class C fly ash is not available. Indeed, the
role of Class C fly ash with respect to sulfate resistance is not clear.13.18
Tests on air-entrained concrete with a water/cement ratio of 0.33 and a Class F fly
ash content of 58 per cent by mass of cementitious material have shown an excellent
resistance to freezing and thawing.13.30 It should be noted that, for concrete exposed to
de-icing agents, ACI 318-0213.116 limits the mass content of fly ash and other
pozzolanas to 25 per cent, in quantities up to 20 per cent of the total mass of
cementitious material, this fly ash has no adverse effect on the resistance to freezing
and thawing of air-entrained concrete. At high contents of Class C fly ash, the
resistance was found to be impaired, possibly due to an increase in the porosity of the
hardened cement paste caused by the movement of fibrous ettringite into the air
voids.13.1
With respect to air entrainment of fly ash concrete, the problems caused by carbon,
discussed on p. 551, should be borne in mind.
Bilodeau et al.13.124 found that fly ash, both Class F and Class C, at least when
present in large proportions, results in concrete with a poor resistance to de-icing
agents, even though the concrete has a good resistance to freezing and thawing. The
reasons for this have not been established.
Because of the reduced permeability of mature concrete containing fly ash, the
chloride ingress into such concrete is reduced. Even when the content of Class F fly
ash is as high as 60 per cent by mass of cementitious material, the passivation of steel
embedded in mortar and the risk of corrosion were found to be unimpaired.13.24 This
was confirmed by other tests on concretes with high fly ash contents (58 per cent of
the total cementitious material) and water/cement ratios between 0.27 and 0.39, which
have shown a very good resistance to chloride penetration.13.24
Nevertheless, in some countries13.12 the use of fly ash in prestressed concrete is not
permitted, it being thought that carbon in the fly ash may contribute to stress
corrosion of the prestressing steel.
The abrasion resistance of concrete containing fly ash, Class F or Class C, is
unimpaired13.29 or possibly even improved.13.31
Fly ash, in adequate quantity in the mix, is beneficial in reducing the alkali–silica
reaction (see p. 523) but the mechanisms involved are complex and imperfectly
understood. The beneficial effects may arise from the denser structure of the hydrated
cement paste which impedes the movement of ions, or from the preferential reaction
of the alkalis with the fly ash so that they are not available for reaction with the silica
in the aggregate.13.28 It should be pointed out that fly ash itself contains alkalis, but
typically only about one-sixth of the total alkali content in the fly ash is water-soluble,
and therefore potentially reactive, the remainder being combined. Whether or not the
fly ash contributes alkalis to the pore water in concrete seems to depend on the
alkalinity of the cement used.13.27
There is no beneficial effect of fly ash with respect to the alkali–carbonate reaction.

717
Concretes containing ground granulated blastfurnace slag (ggbs)
Portland blastfurnace slag cement (see Chapter 2) has been used for more than a
century, although in recent years there has been an increasing use of mixing the
Portland cement and the ground granulated blastfurnace slag (ggbs) components
direct in the concrete mixer. An advantage of this procedure is that the proportion of
Portland cement and ggbs can be varied at will; a concomitant disadvantage is that an
additional silo is required.
Because slag is produced at the same time as pig iron, the production control
ensures a low variability of both materials. The slag is subsequently granulated or
pelletized; for convenience, the term ‘granulated’ is generally used. The granulated
slag can be ground to a fineness of any desired value, but usually greater than 350
m2/kg, that is, finer than Portland cement. Increased fineness leads to increased
activity at early ages, and occasionally ggbs with a fineness in excess of 500 m2/kg is
used.13.34
There are several possible beneficial effects of incorporating ggbs in the mix.
These are: the fresh concrete has an improved workability; the heat development is
slower so that the peak temperature is lower; a denser microstructure of hydrated
cement paste is achieved and this improves long-term strength and, especially,
durability; and the risk of alkali–silica reaction can be eliminated, regardless of the
alkali content of the Portland cement or the reactivity of the aggregate.13.69
The choice of the fineness of ggbs and of its content in the total cementitious
material depend on the purpose of the use of ggbs in the concrete.
Influence of ggbs on properties of fresh concrete
The presence of ggbs in the mix improves workability and makes the mix more
mobile but cohesive. This is the consequence of a better dispersion of the
cementitious particles and of the surface characteristics of the ggbs particles, which
are smooth and absorb little water during mixing.13.32 However, the workability of
concrete containing ggbs is more sensitive to variations in the water content of the
mix than is the case with Portland-cement-only concrete. When ground to a high
fineness, ggbs reduces bleeding of concrete.
Mixes containing ggbs have been sometimes found to exhibit an early loss of
slump, but there are also reports of a low rate of slump loss.13.32
The presence of ggbs in the mix leads to retardation at normal temperatures,
typically 30 to 60 minutes.13.32
Hydration and strength development of concrete containing ggbs
Because a blend of Portland cement and ggbs contains more silica and less lime than
Portland cement alone, hydration of the blended cement produces more C-S-H and
less lime than Portland cement alone. The resulting microstructure of the hydrated
cement paste is dense. However, the initial hydration of ggbs is very slow because it
depends upon the breakdown of the glass by the hydroxyl ions released during the
hydration of the Portland cement. In a manner similar to blended cements containing
pozzolanas, reaction of ggbs with calcium hydroxide takes place.
The progressive release of alkalis by the ggbs, together with the formation of
calcium hydroxide by Portland cement, results in a continuing reaction of ggbs over a
long period. Thus, there is a long-term gain in strength13.132 (see Fig. 13.3). As an
example, Roy13.9 quoted that 8 to 16 per cent of ggbs has hydrated at 32 days, and 30 to

718
37 per cent at 28 days. However, the later rate of hydration of blended cement
containing ggbs is accelerated. Thus, overall, the peak temperature of concrete caused
by hydration of cement is reduced by the inclusion of ggbs in the mix.

Fig. 13.3. Development of compressive strength of concrete (measured on cubes)

moist cured at room temperature for various contents of ggbs by mass of total
cementitious material13.132 (Copyright ASTM–reproduced with permission)
The solubility of alkali hydroxides increases with an increase in temperature.
Consequently, the reactivity of ggbs at higher temperatures is considerably increased.
Steam curing of concrete containing ggbs can, therefore, be used.13.123 Moreover, the
harmful effects of high early temperature upon long-term strength and permeability
are less pronounced in concrete containing ggbs than in Portland-cement-only
concrete.13.2,13.33 Conversely, at temperatures below about 10 °C (50 °F) the strength
development is poor13.42 and the use of ggbs is undesirable.
Greater fineness of ggbs leads to a better strength development, but only at later
ages, because activation of ggbs must first take place. A greater fineness of Portland
cement speeds up the activation.
Other factors influencing the reactivity of ggbs are the chemical composition of the
slag (see p. 80), and the glass content. However, attempts to relate the reactivity of the
slag to its chemical composition by a single ‘chemical modulus’ or ‘hydraulic index’
have not proved successful. While a high glass content is essential, a few per cent of
crystalline material may be beneficial with respect to the reactivity of ggbs because
these crystals act as nuclei for hydration.13.125 An important factor is the concentration
of the alkalis in the total cementitious material; thus, the properties of the Portland

719
cement used with a given ggbs are a factor. Generally, a better development of
strength is found with finer cements and with cements that have high contents of C3A
and of the alkalis.13.96
The proportions of ggbs and Portland cement influence the development of
strength of the resulting concrete. For the highest medium-term strength, the
proportions are about 1:1, that is, a 50 per cent content of ggbs in the cementitious
material;13.123 the early strength is inevitably lower than with the same content of
cementitious material consisting of Portland cement only. In many structures,
however, the early strength is not important. An example of the development of
strength of mortars containing varying proportions of ggbs is shown in Fig.
13.4 which suggests an optimum ggbs content of about 50 per cent from a strength
standpoint.13.36 Very good development of strength of concretes containing 50 to 75 per
cent of ggbs, with a total content of cementitious material between 300 and 420
kg/m3 (300 to 500 lb/yd3) has been reported.13.35

Fig. 13.4. Influence of content of ggbs in the total cementitious material (by mass)
on the strength of mortar at various ages13.36
Reference to the beneficial effects of higher temperatures upon the strength of
concrete containing ggbs was made earlier in this section. In this connection, it should
be noted that tests comparing the development of strength of concretes with and
without ggbs, using specimens cured under standard conditions of temperature, may
not give a correct picture. In actual structural members, the temperature is likely to

720
rise in consequence of the initial hydration of Portland cement so that the strength
development would be greater than in standard test specimens.13.69
Prolonged moist curing of concrete containing ggbs is particularly important
because the initial low rate of hydration results in a system of capillary pores which
allows the loss of water under drying conditions. If this happens, continuing hydration
cannot take place. Japanese recommendations for curing may be of interest; these are
shown in Table 13.2.

Table 13.2. Japanese Recommendations for Moist Curing of Concrete

Containing Different Percentages (by Mass of Total Cementitious Material) of
Ground Granulated Blastfurnace Slag13.42

The incorporation of ggbs in concrete does not alter significantly the usual
relations between compressive strength and fiexural strength or between the
compressive strength and the modulus of elasticity.13.42 Occasional differences have
been reported, but the assumption of any special relation has to be based on tests.
Shrinkage of concrete containing ggbs is initially increased13.123 but, overall, shrinkage
and creep are not adversely affected by the use of ggbs.13.42
A comment on the colour of concrete containing ggbs may be of interest. The ggbs
itself is lighter in colour than Portland cement, and this is reflected in the colour of the
resulting concrete, especially at high contents of ggbs. There is an additional effect:
several days after placing, the concrete may acquire a bluish hue due to the reactions
of iron sulfide in the slag. Upon subsequent oxidation of the sulfide, usually over a
period of several weeks, the bluish tinge disappears. However, if the concrete is
sealed early or remains wet, the oxidation may be prevented.13.42
Durability aspects of concrete containing ggbs
Tests on mortar containing ggbs have shown that its water permeability is reduced by
a factor of up to 100.13.43 There is also a very large reduction in diffusivity of mortar
containing ggbs, especially with respect to chloride ions.13.43
Tests on concrete containing ggbs have confirmed good resistance to penetration
by chloride ions.13.35 Daube and Bakker13.126 have shown that, when the ggbs content is at
least 60 per cent by mass of the cementitious material and the water/cement ratio is
0.50, the diffusion coefficient of the concrete exposed to chloride ions is at least ten
times smaller than when the cementitious material consists entirely of Portland
cement.

721
 The beneficial effects of ggbs arise from the denser microstructure of hydrated
cement paste, more of the pore space being filled with C-S-H than in Portland-
cement-only paste.
As a result of the improved microstructure of hydrated paste of a Portland cement–
ggbs blend, and also because of a low content of calcium hydroxide, the resistance to
sulfate attack is improved. Hooton and Emery13.128 reported that blended cement
containing 50 per cent by mass of ggbs (with 7 per cent of A12O3) and Type I Portland
cement (with a C3A content of 12 per cent) exhibits the same sulfate resistance as
sulfate-resisting (Type V) cement when tested in mortar. To be effective, the content
of ggbs must be at least 50 per cent by mass of the total cementitious material, and
preferably 60 to 70 per cent.
The very low penetrability of concrete which contains ggbs is effective also in
controlling the alkali–silica reaction: the mobility of the alkalis is greatly reduced.
This effect is complemented by the incorporation of the alkalis in the products of
reaction of ggbs, especially at higher temperatures.13.36 The beneficial effects of ggbs
when used in conjunction with siliceous aggregates suspected of alkali reactivity or
with Portland cement with an alkali content up to 1.0 per cent are of great importance.
The situation with respect to freezing and thawing is different. Concrete of
appropriate mix proportions containing ggbs has the same resistance to freezing and
thawing as concrete made with Portland cement only. However, the inclusion of ggbs
in air-entrained concrete has no beneficial effect.13.32,13.123 In view of the beneficial
influence of ggbs on the permeability of the resulting concrete, it is not clear why the
inclusion of ggbs in concrete does not improve its resistance to freezing and thawing
in a manner similar to the beneficial effects of a reduced water/cement ratio. In this
connection, it is relevant to note that, for concrete exposed to de-icing agents, ACI
318-0813.116 imposes a limit on the content of ggbs of 50 per cent of the total
cementitious material. When both ggbs and fly ash are included in the mix, the mass
of both these materials together is limited to 50 per cent of the total mass of
cementitious material; the limitation on fly ash alone of 25 per cent (see p. 742)
applies also when it is used with ggbs.
It should be pointed out that, in order to achieve the same resistance to freezing
and thawing with ggbs in the mix as is offered by Portland-cement-only concrete,
prolonged moist curing prior to exposure to freezing and thawing is essential.
A beneficial effect of the inclusion of ggbs in the mix upon the resistance of
concrete to de-icing salt scaling has been reported by Virtanen,13.37 but this has not
been confirmed.
With respect to carbonation, the effects of ggbs are two-fold. Because of the small
amount of calcium hydroxide present in the hydrated cement paste, carbon dioxide
does not become fixed near the surface of the concrete so that there is no pore-
blocking formation of calcium carbonate. Consequently, at early ages, the depth of
carbonation is significantly greater than in concrete containing Portland cement
only.13.34 On the other hand, the low permeability of well-cured concrete containing
ggbs prevents a continuing increase in the depth of carbonation.13.37,13.43 For this reason,
except when the ggbs content is very high, there is no increased risk of corrosion of
steel reinforcement through a reduction in the alkalinity of the hydrated cement paste
and depassivation of the steel.13.32
Concrete containing silica fume

722
The physical properties of silica fume were described in Chapter 2. The use of this
cementitious material continues to increase despite its relatively high cost. Silica fume
is particularly valued in making high performance concrete, which will be discussed
later in this chapter. In the present section, the general features of the use of silica
fume in concrete will be considered. It can be noted that there exists no British
Standard on silica fume and the ACI guide for the use of silica fume in concrete, ACI
234R-9613.159 was first published in 1996.
The very high reactivity of silica fume with calcium hydroxide produced by the
hydration of Portland cement was mentioned in Chapter 2. Because of this reactivity,
it is possible to use silica fume as a replacement for a small proportion of Portland
cement. This is done on the basis of 1 part of silica fume instead of 4 or even 5 parts
of Portland cement by mass; a maximum of 3 to 5 per cent of silica fume is
used.13.40 When using this approach for low- or medium-strength concretes, the strength
is unaffected by the silica fume replacement. Because, in such concretes, the
water/cement ratio is high or medium, the use of superplasticizers is not necessary.
Additional benefits of replacement by silica fume are: reduced bleeding and improved
cohesion of the mix. However, such use of silica fume is limited to some geographic
areas with an abundant local supply of silica fume which can be used in the low bulk-
density form (see p. 87).
By far the largest use of silica fume is for the purpose of producing concrete with
enhanced properties, mainly high early strength or low penetrability. The beneficial
effects of silica fume are not limited to its pozzolanic reaction: there is also a physical
effect of the ability of the extremely fine particles of silica fume to be located in very
close proximity to the aggregate particles, that is, at the aggregate–cement paste
interface. This zone is known to be a source of weakness in concrete, the reason being
the wall effect which prevents the particles of Portland cement from packing tightly
against the surface of the aggregate. Such packing is achieved by the particles of silica
fume, which are, typically, 100 times smaller than cement particles. A contributing
factor is the fact that silica fume, because of its high fineness, reduces bleeding so that
no bleed water is trapped beneath coarse aggregate particles. In consequence, the
porosity in the interface zone is reduced, compared with a mix not containing silica
fume. Subsequent chemical reaction of silica fume results in a still lower porosity in
the interface zone which, in consequence, is no longer particularly weak, either in
terms of strength or of permeability.
The above argument explains why too low a content of silica fume, say below 5
per cent of the total mass of cementitious material, does not lead to a high strength of
concrete: the volume of silica fume is inadequate to cover the surface of all coarse
aggregate particles. It is also evident that a large volume of silica fume is only
marginally more beneficial than about 10 per cent because the excess silica fume
cannot be located at the surface of the aggregate. It is useful to point out that the
beneficial effect of changes in the hardened cement paste in the interface zone cannot
exist in neat cement paste because, in the absence of aggregate, there is no interface
zone; this was confirmed by Scrivener et al.13.5
Influence of silica fume on properties of fresh concrete
It is essential that silica fume be thoroughly and uniformly dispersed in the mix. For
this reason, the mixing time should be extended, and especially so with silica fume in
the densified form of micropellets. The sequence of feeding materials into the mixer is
of importance and is best established by trial and error.

723
 The very large surface area of the particles of silica fume, which have to be wetted,
increases the water demand, so that, in mixes with a low water/cement ratio, it is
necessary to use a superplasticizer. In this way, it is possible to maintain both the
required water/cement ratio and the necessary workability.
The effectiveness of superplasticizers is enhanced by the presence of silica fume.
For instance, in mixes with a slump of 120 mm, a given dosage of a superplasticizer
was found to reduce the water demand by 10 kg/m3 in a Portland-cement-only
concrete. The same dosage maintained the slump when the silica fume content was 10
per cent by mass of cementitious material. Without the superplasticizer, the water
demand due to the inclusion of silica fume in the mix13.122 would have risen by 40 kg/m3.
It can, therefore, be seen that the use of both silica fume and a suitable
superplasticizer is beneficial and makes it possible to use low water/cement ratios at a
given workability.13.39 The lower water/cement ratio results in an increase in strength
which is larger than would be expected solely from the pozzolanic action of silica
fume. However, in relative terms, the effect of the lower water/cement ratio upon
strength is smaller than the overall direct effect of silica fume.13.5
At this stage, it may be useful to note that the pattern of the relation between
compressive strength and the water/cementitious material ratio is the same for
concretes with and without silica fume but, at the same ratio, concrete with silica
fume has a higher strength. Examples of the relation between the 28-day compressive
strength of 100 mm cubes and the water/cementitious material ratio for concretes with
8 and 16 per cent of silica fume by mass of the total cementitious material, are shown
in Fig. 13.5. The same figure also shows the relation for concrete containing Portland
cement only.13.62

724
 Fig. 13.5. Relation between the compressive strength (measured on 100 mm
cubes) and the water/cement ratio for concretes with different contents of silica
fume (by mass of total cementitious material)13.62
The presence of silica fume affects significantly the properties of fresh concrete.
The mix is strongly cohesive and, in consequence, there is very little bleeding, or even
none. The reduced bleeding can lead to plastic shrinkage cracking under drying
conditions, unless preventive measures are taken. On the other hand, voids caused by
trapped bleed water are absent.
The cohesive character of the mix affects the slump so that, for both mixes equally
to be capable of compaction, the mix with silica fume needs a slump 25 to 50 mm (1
to 2 in.) higher than a mix containing Portland cement only.13.55,13.57 Mixes with a very
high content of cementitious material tend to be ‘sticky’ and do not easily allow the
slump cone to be lifted. It has, therefore, been suggested that the slump test is

725
inappropriate and that the flow test is preferable.13.38 The ‘sticky’ nature should not be
misinterpreted: as soon as vibration is applied, the mix becomes ‘mobile’. However,
in order to avoid an excessively sticky mix, it is recommended13.99 that the water
content should not be lower than 150 kg/m3 (250 lb/yd3) when the fine aggregate is
angular in shape, or not lower than 130 kg/m3 (220 lb/yd3) when rounded fine
aggregate is used.
The cohesiveness of concrete containing silica fume makes it satisfactory for
pumping and for underwater concreting, as well as for use as flowing concrete13.55 (see
p. 259). Entrained air remains stable,13.57 but an increased dosage of the air-entraining
admixture is required because of the high fineness of silica fume. In addition, there
are problems in obtaining a suitable air-void system when superplasticizers are used
(which is usually the case with silica fume mixes).
There are no reports of incompatibility of silica fume with admixtures in general. It
is useful to observe that the retarding effect of lignosulfonate-based admixtures is
smaller when silica fume is present in the mix. Consequently, larger dosages of these
admixtures can be used without causing an excessive retardation.13.55
Hydration and strength development of the Portland cement–silica fume system
In addition to the pozzolanic reaction between the amorphous silica in silica fume and
calcium hydroxide produced by the hydration of Portland cement, silica fume
contributes to the progress of hydration of the latter material. This contribution arises
from the extreme fineness of the silica fume particles which provide nucleation sites
for calcium hydroxide. Thus, early strength development takes place.
Silica fume dissolves in a saturated solution of calcium hydroxide within a few
minutes.13.9 Therefore, as soon as enough Portland cement has hydrated to result in
saturation of the pore water with calcium hydroxide, calcium silicate hydrate is
formed on the surface of the silica fume particles. This reaction proceeds, initially, at
a high rate. For example, when the mass of silica fume was 10 per cent of the total
mass of cementitious material, one-half of the silica fume was observed to react in 1
day, and two-thirds during the first 3 days. However, subsequent reaction was very
slow, only three-quarters of the silica having hydrated at 90 days.13.8
The acceleration of hydration processes by silica fume occurs also when, in
addition to Portland cement, ggbs is present in the mix.13.46
A consequence of the rapidity of the early reactions in concretes containing silica
fume is that the development of heat of hydration in such concretes may be as high as
when rapid-hardening Portland (Type III) cement is used alone.13.9
The behaviour of concrete with silica fume beyond the age of about 3 months
depends on the moisture conditions under which the concrete is stored. Up to the age
of years, tests showed a small increase in compressive strength of wet-stored
concretes with 10 per cent of silica fume (by mass of total cementitious material) and
water/cement ratios of 0.25, 0.30, and 0.40.13.58 Under dry storage conditions,
retrogression of strength, typically up to 12 per cent below the peak value at about 3
months, was observed in tests on laboratory specimens.13.58 However, the strength of
concrete containing silica fume, determined on cores up to 10 years old, clearly shows
no retrogression of strength.13.47 This finding is of importance because the behaviour of
test specimens in which moisture gradients exist may be misleading.13.56
The C-S-H produced by silica fume has a lower C:S ratio than the C-S-H resulting
from the hydration of Portland cement alone. Values of the C:S ratio in the products

726
of hydration of silica fume have been found13.9 to be as low as 1. The C:S ratio is lower
at high contents of silica fume in the cementitious material.13.41
One consequence of the high early reactivity of silica fume is that the mix water is
rapidly used up; in other words, self-desiccation takes place.13.49 At the same time, the
dense microstructure of the hydrated cement paste makes it difficult for water from
outside, if available, to penetrate toward the unhydrated remnants of Portland cement
or silica fume particles. In consequence, strength development ceases much earlier
than with Portland cement alone; some experimental data are shown in Table
13.313.49 from which it can be seen that there was no increase in strength beyond 56
days. The data of Table 13.3 refer to mixes with a total content of cementitious
material of 400 kg/m3, sulfate-resisting Portland (Type V), cement, silica fume
contents of 10, 15, and 20 per cent by mass of total cementitious material, and a
water/cement ratio of 0.36; the concrete specimens were maintained under moist
conditions.

Table 13.3. Strength Development of Test Cylinders of Concretes Containing

Silica Fume13.49

The contribution of silica fume to the early strength development (up to about 7
days) is probably through improvement in packing, that is, action as a filler and
improvement of the interface zone with the aggregate.13.45 The bond of the hydrated
cement paste with aggregate, especially the larger particles, is greatly
improved,13.50 allowing the aggregate better to participate in stress transfer. Some
contrary arguments about the role of silica fume have been advanced,13.44 but they are
likely to reflect specific test conditions rather than intrinsic behaviour.
The contribution of a given amount of silica fume to the strength of concrete
arising from packing and interface effects should remain constant with time. This is
unlike the effect of pozzolanic activity which continues to take place. Indeed, at a
fixed content of silica fume, the increase in strength of concrete between 7 and 28

727
days was found to be independent of the value of the 7-day strength.13.59 The
contribution of silica fume to strength at, say, 28 days should, however, increase with
an increase in the content of silica fume in the mix (up to a certain limit). This was
found to be the case for concretes with 28-day strengths between approximately 20
and 80 MPa (3000 to 12 000 psi): the increase in strength was 7 MPa for a 10 per cent
content, and 16 MPa for a 20 per cent content of silica fume.13.59
The relation between the content of silica fume in the mix and the resulting
strength, just mentioned, has encouraged numerous attempts to devise a so-called
efficiency factor of silica fume for strength. Other efficiency factors have been
derived for other properties of concrete made with silica fume, such as
permeability.13.55 The various factors differ from one another. For this reason, and also
because the effects of silica fume are influenced by the properties of the Portland
cement used, the ‘efficiency factor’ approach is not considered to have sufficient
validity.
The continuing pozzolanic activity of silica fume results in a reduction in the pore
size in the hydrated cement paste. Test data demonstrating the existence of very small
pores in hydrated paste of blended sulfate-resisting cement (Type V) and silica fume
are shown in Table 13.4; mercury intrusion porosimetry was used. The same table
shows that the reduction in total porosity of hydrated cement pastes containing silica
fume is small, as compared with a paste made with a sulfate-resisting (Type V)
cement only.13.49 It can be seen, thus, that the main effect of silica fume is to reduce the
permeability of the hydrated cement paste, but not necessarily its total porosity.
Whereas the presence of silica fume representing 10 per cent by mass of the total
cementitious material has a large effect on the pore system, increasing further content
of silica fume in the cementitious material leads only to a small change. This accords
with the earlier observations (see p. 669) to the effect that there is no beneficial effect
of the presence of particles of silica fume in excess of those required to cover the
surface of the aggregate and to infill the space between the particles of Portland
cement.

Table 13.4. Pore Characteristics of Mortars Containing Sulfate-Resisting

Cement and Silica Fume13.49

728
 As is the case with all pozzolanic reactions, prolonged moist curing of concrete
containing silica fume is necessary, especially because of its strength contribution
between the ages of 3 and 28 days.13.55 Surprisingly, tests on mortar containing silica
fume have shown that the beneficial effects of a prolonged moist curing on flexural
strength are much smaller than on the compressive strength.13.89 Confirmation, or
otherwise, of similar behaviour in concrete is not available. Differences in curing
apart, the relation between the tensile or flexural strength and compressive strength is
not affected by the presence of silica fume in the concrete.13.55,13.99
The modulus of elasticity of concrete containing silica fume is somewhat higher
than is the case with Portland-cement-only concretes of similar strength.13.55 It has been
reported that concrete containing silica fume is more brittle, but this has not been
confirmed.13.55
Durability of concrete containing silica fume
In the preceding section, we discussed the importance of adequate curing of concrete
containing silica fume from the standpoint of reactions of hydration. As far as
durability is concerned, we should note that a consequence of more advanced
hydration is a reduced permeability; as already mentioned, adequate curing is of
particular importance. Generally, for concretes of equal strength, the reduction in
permeability due to a longer period of curing is greater in concrete containing silica
fume than in Portland-cement-only concrete.13.127
The desirable minimum curing period depends, among other things, on
temperature, which in the field may be subject to considerable variation. Low
temperature slows down the hydration reactions involving silica fume even more than
is the case in Portland-cement-only concretes.13.55 However, upon a subsequent rise in

729
temperature, the usual reactions take place13.121 and the accelerating effect of a higher
temperature is greater than in the case of Portland cement alone.13.55 Also, the harmful
effects of a higher temperature on pore structure are smaller in the presence of silica
fume.13.127
It is important to note that carbonation is particularly adversely affected by
inadequate curing.13.55
The influence of silica fume upon the permeability of concrete is greater than is
indicated by tests on hydrated cement paste because, in the former case, silica fume
reduces the permeability of the transition zone around the aggregate particles, as well
as the permeability of the bulk paste.13.57 The influence of silica fume upon
permeability of concrete is very large: a 5 per cent content of silica fume was reported
by Khayat and Aïtcin13.57 to reduce the coefficient of permeability by 3 orders of
magnitude. Thus, in relative terms, the influence of silica fume upon permeability is
much larger than upon compressive strength.
A consequence of reduced permeability is a greater resistance to the ingress of
chloride ions. Even using Portland cements with C3A contents up to as much as 14 per
cent, the presence of 5 to 10 per cent of silica fume in the total cementitious material
greatly slows down the ingress of chloride ions into concrete.13.48,13.138 ACI 318-
0813.116 limits the content of silica fume to 10 per cent when the concrete is to be
exposed to de-icing agents. The reduction in the diffusivity of chlorides due to the
presence of silica fume in hydrated cement paste is larger at water/cementitious
material ratios greater than 0.4 than at extremely low values of the water/cementitious
material ratio;13.51 in the latter case, the hydrated cement paste has a very low
diffusivity, even without silica fume.
The sulfate resistance of concrete containing silica fume is good, partly because of
a lower permeability, and partly in consequence of a lower content of calcium
hydroxide and of alumina, which have become incorporated in C-S-H. Tests on
mortar have shown also the beneficial effect of silica fume upon resistance to
solutions of magnesium, sodium, and calcium chlorides.13.52 The role of pozzolanas in
controlling expansive alkali–silica reaction was discussed on p. 520. Silica fume is
particularly effective in this respect.13.53 It can be added that a consequence of the lower
C:S ratio of the products of reaction of silica fume is an increased ability of these
products to incorporate ions such as the alkalis or aluminium.13.55
With respect to resistance to freezing and thawing, some investigators13.61 reported a
poor resistance of air-entrained concrete containing silica fume as compared with
Portland-cement-only concrete. A possible explanation is that, with an adequate
entrained-air content, the concrete containing silica fume had a larger air-void spacing
factor and, at the same time, the dense structure of the hydrated cement paste
prevented the movement of water. On the other hand, other investigators13.60 found a
good resistance of concretes containing silica fume to freezing and thawing, and also
to scaling by de-icing agents. Experience with structures in situ gave variable
results.13.37
Resolution of this conflict in reports on performance would require a detailed
knowledge of test procedures used, including the maturity and moisture condition of
the concretes at the time of the test. Indeed, the influence of silica fume upon
resistance to freezing and thawing is complex. After a period of moist curing, the pore
size in the hydrated cement paste becomes smaller (see p. 658); in consequence, the
freezing point of pore water is reduced (see p. 539). In the interior of concrete, self-

730
desiccation is likely to have reduced the water content below the critical level of
saturation so that freezing would not cause damage. The fine pore system also makes
it difficult for the concrete to become re-saturated after drying.13.88 On the other hand, a
dense paste with a very low permeability does not allow a rapid enough movement of
water out of pores subjected to freezing and into an air void. Thus, rapid freezing
would lead to damage.13.57
The preceding discussion shows that generalizations about the influence of silica
fume on the resistance of concrete to freezing and thawing, and even more so to
scaling by de-icing agents, are not possible: much depends on the particular concrete
used, on its treatment prior to freezing and thawing, and on the rapidity of temperature
changes. It is, therefore, not surprising that many publications present conflicting
results, and there would be little value in reviewing them in this book. For practical
purposes, the only conclusion which can be drawn is that it is necessary to test any
concrete which it is proposed to use, and the test results have to be interpreted in the
light of the expected conditions of exposure.
Because silica fume reduces the alkali content in the pore water, the pH of pore
water becomes lowered. Tests on mature cement pastes made with Portland cement
with a very high alkalinity (pH of 13.9) have shown a reduction in the value of pH
caused by the inclusion in the mix of 10 per cent of silica fume to be 0.5; 20 per cent
of silica fume reduced the value of pH by 1.0.13.139 Even with the last-named reduction,
the value of pH was 12.9. Havdahl and Justnes13.129 confirmed that the pH stays above
12.5. Thus, the alkalinity is adequately high for the protection of reinforcing steel
from corrosion.13.55
The presence of silica fume in concrete has a beneficial influence upon resistance
to abrasion because, in the absence of bleeding, no weak top layer is formed and also
because of a better bond between the hydrated cement paste and the coarse aggregate;
differential wear and loosening of particles do not, therefore, occur.13.57
Shrinkage of concrete containing silica fume is somewhat larger, typically 15 per
cent, than in Portland-cement-only concrete.13.49
The darker colour of some silica fumes was mentioned on p. 85. This influences
the colour of the resulting concrete. However, the colour becomes lighter after a few
weeks, but the reasons for this are not clear.13.55
High performance concrete
High performance concrete is not a revolutionary material, nor does it contain
ingredients which are not used in the concrete considered this far. Rather, high
performance concrete is a development of the concretes discussed in the last few
sections.
The very name ‘high performance concrete’ smacks of advertising an allegedly
distinct product. A former name was ‘high strength concrete’ but, in many cases, it is
a high durability that is the required property, although, in others, it is high strength,
either very early, or at 28 days, or even later. In some applications, a high modulus of
elasticity is the property sought.
With respect to strength, we should note that the meaning of the term ‘high
strength’ has changed significantly over the years: at one time, 40 MPa (or 6000 psi)
was considered high; later on, 60 MPa (or 9000 psi) became viewed as high strength
concrete. In this book, high performance in terms of strength will be taken as a
compressive strength in excess of 80 MPa (or 12 000 psi). In passing, it can be said

731
that, at these high strengths, the difference between test results on cubes and on
cylinders is minimal, so that, except for compliance purposes, the distinction between
the two types of test specimens is of little importance. Testing high performance
concrete is discussed on p. 685.
One more comment about nomenclature may be in order. In some publications, a
subdivision of high performance concrete into classes according to strength is
introduced and terms such as ‘very high performance concrete’ are used. This does
not seem to be a rational approach to a material with a continuous gradation in
properties and with no discontinuities in ingredients.
High performance concrete contains the following ingredients: common, albeit
good quality, aggregate; ordinary Portland (Type I) cement (although rapid-hardening
Portland (Type III) cement can be used when high early strength is required) at a very
high content, 450 to 550 kg/m3 (760 to 930 lb/yd3); silica fume, generally 5 to 15 per
cent by mass of the total cementitious material; sometimes, other cementitious
materials such as fly ash or ground granulated blastfurnace slag; and always a
superplasticizer. The dosage of the superplasticizer is high: 5 to 15 litres per cubic
metre of concrete, depending on the solids content in the superplasticizer, as well as
on its nature. Such a dosage allows a reduction in water content of about 45 to 75
kg/m3 of concrete.13.79 Other admixtures can also be present, but polymers, epoxies,
fibres, and processed aggregates such as calcined bauxite sand are excluded from
consideration in this book. It is essential that high performance concrete is capable of
being placed in the structure by conventional methods and that it is cured in the usual
manner, although particularly good moist curing is required. What makes the concrete
a high performance one is a very low water/cement ratio: always below 0.35, often
around 0.25, and occasionally even 0.20.
The above discussion makes it clear that what was said earlier in this chapter about
the properties of concrete containing silica fume and a superplasticizer applies to high
performance concrete but, because of the very low water/cement ratio in the latter,
these properties are accentuated. Indeed, high performance concrete can be said to be
a logical development of concrete containing silica fume and a superplasticizer. To
give an example, it is possible to produce a mix with a slump of 180 to 200 mm (7 to
8 in.) at a water/cement ratio between 0.2 and 0.3; thus, the water content is 130 to
140 kg per cubic metre of concrete (220 to 240 lb/yd3), as compared with 170 to 200
kg/m3 (290 to 340 lb/yd3) in a non-air-entrained mix of ordinary concrete with a slump
of 100 to 120 mm (4 to 5 in.).
It was mentioned earlier that high performance concrete means concrete with a
high strength or a low permeability. The two properties, although not necessarily
concomitant, are linked to one another because high strength requires a low volume of
pores, especially of the larger capillary pores. The only way to have a low volume of
pores is for the mix to contain particles graded down to the finest size: this is achieved
by the use of silica fume which fills the spaces between the cement particles and
between the aggregate and the cement particles. However, the mix must be
sufficiently workable for the solids to be dispersed in such a manner that dense
packing is achieved, which requires deflocculation of cement particles. This is
achieved by the use of a superplasticizer at a large dosage. The superplasticizer must
be effective with the given Portland cement, that is, the two materials must be
compatible.
When the above conditions have been satisfied, high performance concrete is
achieved. The concrete is very dense, it has a minimal volume of capillary pores, and

732
these pores become segmented upon curing. At the same time, a significant proportion
of Portland cement remains unhydrated, even when the concrete is in contact with
water because water cannot penetrate through the pore system so as to reach the
unhydrated remnants of Portland cement. These remnants can be viewed as very fine
‘aggregate’ particles which are extremely well bonded to the products of hydration.
Properties of aggregate in high performance concrete
Although common aggregates are used in making high performance concrete, in
concretes of very high strength, the strength of the coarse aggregate particles
themselves can be critical. In consequence, the strength of the parent rock is of
importance, but the bond strength of the aggregate particles can also be a limiting
factor.13.91 The mineralogical characteristics of coarse aggregate have been found to
influence the strength of the resulting concrete, but no simple guidance on the
selection of aggregate is available.13.64
The criterion of the strength of aggregate is valid when a high long-term strength
of concrete is required. If, however, the desired property of high performance
concrete is a high strength at a very early age (say, 40 MPa at 2 days) but a higher
strength in the long term is unnecessary, then the strength of the aggregate particles is
unimportant.
Generally, however, good quality aggregate must be used. To ensure good bond
between the coarse aggregate particles and the matrix, these particles should be
approximately equi-dimensional.13.78 It should be remembered that the shape of crushed
particles depends, in addition to the parent rock type and its bedding, also on the
method of crushing used, impact crushers generally producing few elongated or flaky
particles. Gravel is satisfactory as far as shape is concerned and it can be used in high
performance concrete,13.78 but the aggregate–matrix bond may be inadequate when the
surface texture of the gravel is very smooth.
Cleanliness of the aggregate, absence of adhering dust, and uniformity of grading
are essential. Durability of coarse aggregate particles is vital when the concrete
containing the given aggregate is likely to be exposed to freezing and thawing.
Fine aggregate should be rounded and uniformly graded, but rather coarse, because
the rich mixes used in high performance concrete have a high content of fine particles;
a fineness modulus of between 2.8 and 3.2 is sometimes recommended.13.131 However,
experience with high performance concrete in terms of the range of aggregates used,
both fine and coarse, is limited to only a few geographical areas so that
generalizations are not available.
One more comment should be made about the system of solid particles in the mix.
At the coarse end, large particles of aggregate are undesirable because they introduce
a heterogeneity in the system in that, at the interface, there may be an incompatibility
between the aggregate and the surrounding hydrated cement paste in terms of the
modulus of elasticity, Poisson’s ratio, shrinkage, creep, and thermal properties. This
incompatibility may lead to more microcracking than when the maximum size of

aggregate is smaller than 10 or 12 mm ( or in.). Although a smaller maximum size

of aggregate leads to a higher water demand, this is unimportant when the dosage of
superplasticizer is high so that the water content of the mix is low.
The larger total surface area of the aggregate with a smaller maximum size also
means that the bond stress is lower so that bond failure does not occur. Consequently,

733
in compression tests, failure occurs through the coarse aggregate particles, as well as
through the hydrated cement paste. Development of cracks through the coarse
aggregate particles was observed also in flexural tests on high performance
concrete.13.70 This behaviour means that the bond strength is no lower than the tensile
strength of the aggregate.
The influence of the modulus of elasticity of the coarse aggregate upon the
strength of high performance concrete has not been established, but it is arguable that,
because of the monolithic behaviour of the concrete, aggregate with a low modulus of
elasticity (that is, a modulus not very different from the modulus of elasticity of
hydrated cement paste) leads to lower bond stresses with the matrix. This may be
beneficial with respect to high performance concrete.
Aspects of high performance concrete in the fresh state
The particular proportions of the ingredients of high performance concrete, namely,
the very high cement content, the very low water content, and the high dosage of
superplasticizer, influence the properties of the fresh concrete in some respects in a
manner different from the usual mixes.
First of all, batching and mixing require particular care. Because of the importance
of thorough mixing, using the mixer at less than its rated capacity may be beneficial; a
reduction of one-third, or even one-half, may be desirable.13.98 A longer mixing time
than usual is required to ensure homogeneity of what is usually rather a sticky mix: 90
seconds has been recommended,13.93 but even longer periods may be desirable.
The sequence of feeding the ingredients into the mixer is best established by trial-
and-error, and it can be complicated. In one case, some water and one-half of a
superplasticizer were fed first; then, aggregate and cement; finally, the remainder of
the water and the superplasticizer. Often, a part of the superplasticizer is added only
immediately prior to the placing of concrete. An example of the effect of the mixing
sequence upon slump loss of concrete with a water/cement ratio of 0.25, mixed during
225 seconds is shown in Fig. 13.6.13.81 Three sequences were used: (A) feeding all the
ingredients simultaneously; (B) mixing cement and water prior to the feeding of the
remaining ingredients; and (C) mixing cement and fine aggregate prior to the feeding
of the remaining ingredients. Method A resulted in the lowest slump loss, but this
observation may not be generally valid.

734
 Fig. 13.6. Effect of sequence of batching on slump loss with time since mixing of
concrete with a water/cement ratio of 0.25 and a superplasticizer13.81
To optimize setting time and the development of early strength of high
performance concrete, a combination of a superplasticizer and a lignosulfonate-based
water-reducing admixture or a retarder can be used.13.55
Some of the superplasticizer must be introduced early into the mixer in order to
achieve adequate workability in the first place. The timing of adding the final portion
of the superplasticizer is of particular importance. It is essential to ensure that the
superplasticizer does not become fixed by C3A in the Portland cement and,

735
consequently, is no longer available to maintain a high workability. Such fixing would
occur if SO4–– from the calcium sulfate in Portland cement is not liberated fast enough
to react with C3A. It is, therefore, important to avoid the reaction between the
superplasticizer and C3A by ensuring a compatibility between the superplasticizer and
the Portland cement to be used; this topic is discussed in the next section.
At this stage, one more comment is pertinent. The water requirement is influenced
by the carbon content of the silica fume used; a high content can be detected simply
by a dark colour of the silica fume.13.68
Compatibility of Portland cement and superplasticizer
In the preceding section, we pointed out the difficulty of maintaining an adequate
workability if the superplasticizer becomes fixed by C3A in the Portland cement used.
When this happens, the two materials can be said to be incompatible. Conversely, if
the difficulty is absent, the Portland cement and the superplasticizer are said to be
compatible. Although comparability of cement and admixtures is relevant in ordinary
concrete as well, in high performance concrete, the very low water content greatly
magnifies the consequences of a lack of compatibility because of the competition by
the various ingredients for water for surface wetting and early hydration. The rate of
solubility of calcium sulfate is critical when there is less water to accept the sulfate
ions and, at the same time, there is more C3A (because of a high cement content)
whose reaction must be controlled to ensure workability. For these reasons, tests
involving the given materials, but using a water/cement ratio of about 0.5, do not give
information on the behaviour at a water/cement ratio in the vicinity of 0.3.
In essence, the problem is that of the length of time after mixing, before
the ions from Portland cement become available for reaction with C3A, so that
the sulfonate ends of the superplasticizer molecules do not become fixed. The various
forms of calcium sulfate in Portland cement were discussed on p. 17, and it should be
remembered that gypsum, hemihydrate, and anhydrite have different rates of
solubility. The situation is complicated by the fact that the solubility of anhydrite
depends on its structure and origin.
The solubility and the rate of solution of calcium sulfate are affected by the
superplasticizer, both in so far as its type and its dosage are concerned. In the present
state of our knowledge, a translation of these qualitative factors into a prediction of
compatibility is not possible, and an experimental assessment of the rheological
properties of any given combination of Portland cement and superplasticizer is
necessary.
Nevertheless, it can be said that the important factors in compatibility are as
follows.13.79 For the cement, these are the content of C3A and C4AF, the reactivity of
C3A (which depends on its morphological form and on the degree of sulfurization of
the clinker), the content of calcium sulfate, and the final form of the calcium sulfate in
the ground cement (namely, gypsum, hemihydrate, or anhydrite). For the
polysulfonate, the important factors are the molecular chain length, the position of the
sulfonate group in the chain, the counter-ion type (that is, sodium or calcium), and the
presence of residual sulfates, which affect the cement deflocculation properties.
On the basis of these factors, an ideal cement for high performance concrete from
the rheological point of view can be postulated: not too fine (probably up to 400 m2/kg
determined by the Blaine method), with a very low C3A content whose reactivity is
easily controlled by the sulfate ions derived from the solution of the sulfates present in

736
the cement. An ideal polysulfonate superplasticizer should consist of rather long
molecular chains in which, for example, the sulfonate groups occupy the β-position in
a sodium salt condensate of formaldehyde and naphthalene sulfonates. As far as the
content of residual sulfates in the polysulfonate is concerned, this depends on the
content and solubility of the sulfates in the cement with which the superplasticizer is
to be used: what is necessary is an adequate amount of soluble sulfates in the
mixture.13.79
The preceding guidelines make it possible to eliminate inappropriate cements and
superplasticizers. The next step is laboratory testing on a trial-and-error basis of a
number of neat cement pastes containing combinations of different cements and
superplasticizers for the purpose of establishing the best combination from the
rheological point of view.
The effectiveness of a given superplasticizer with a given cement can be tested by
measuring the time taken by a fixed quantity of neat cement paste made with the two
materials to flow through a standard funnel, known as a Marsh flow cone. The time
decreases with an increase in the dosage of superplasticizer up to a saturation point,
beyond which additional superplasticizer is no longer beneficial. The time to empty
the cone becomes longer when the testing of the cement paste is delayed; this is an
indication of a loss of workability. A compatible combination of Portland cement and
superplasticizer exhibits only a small loss between tests at 5 and 60 minutes, and also
exhibits a definite saturation point beyond which additional superplasticizer is of no
benefit (see Fig. 13.7).13.63

737
 Fig. 13.7. Flow time through a Marsh cone as a function of content of
superplasticizer (by mass of solids) in neat cement paste with a water/cement
ratio of 0.35 after 5 and 60 minutes since mixing13.63
Such tests on neat cement paste make it possible to narrow the choice to a few
cements compatible with one or two superplasticizers which are commercially
available. For the final selection of the cement and the superplasticizer, it is necessary
to make tests on a trial concrete mix because only such tests give truly reliable data on
the slump loss and strength gain.
Aspects of hardened high performance concrete
While there exist no standard, or even typical, mix proportions of high performance
concrete, it is useful to present information on several successful mixes; this is given
in Table 13.5. Several of these mixes contain, in addition to Portland cement and
silica fume, other cementitious materials. There is an economic advantage in using
these various cementitious materials, partly because they are cheaper than Portland
cement, but also because they allow a reduction in the dosage of superplasticizer.13.79

Table 13.5. Mix Proportions of Some High Performance Concretes*

738
 A mix of particular interest is Mix E of Table 13.5 which had a water/cement ratio
of 0.25, a total content of cementitious material of 542 kg/m3 of which only 30 per
cent was Portland cement, and 10 per cent silica fume. The compressive strength at 28
days was 114 MPa but it reached 136 MPa at the age of 1 year. It should be
emphasized that this was not a laboratory concrete but it was produced in a ready-
mixed plant.13.79 It is worth adding that commercial production of high performance
concrete necessitates a very strict and consistent quality control.
At the outset of the discussion of high performance concrete, it was said that this
material is simply an extension of the range of the more usual concrete mixes. This is
confirmed by the continuous nature of the broad relation between strength and the

739
water/cement ratio, illustrated in Fig. 13.8. This figure is based on data cited by
Fiorato13.54 for test cylinders cured in a variety of ways and tested at ages from 28 days
onwards; results of tests on zero-slump concrete without silica fume have been
omitted.

Fig. 13.8. Relation between compressive strength and water/cement ratio for test
cylinders of non-air-entrained concretes containing various cementitious
materials tested at ages between 28 and 105 days (based on ref. 13.54)
In the case of a relatively new material, such as high performance concrete, it is
useful to know whether any retrogression of strength occurs. Tests on cores from a
simulated column made of concrete with a 28-day compressive strength of 85 MPa
(12 300 psi) have shown no change in strength after 2 or 4 years.13.74 Reported
retrogression of strength of test cylinders stored under dry conditions between the
ages of 90 days and 4 years can be explained by self-stressing through drying of the
surface zone of the cylinders;13.56 such drying is absent in concrete in structures.
Information on the relation between the modulus of rupture or the splitting tensile
strength and the compressive strength for high performance concrete is not available,
but ACI 363R-9213.91 suggests expressions applicable up to 83 MPa (12 000 psi). There
are indications that, for compressive strengths above about 100 MPa (14 000 psi),
there is no further increase in the modulus of rupture or in the tensile strength.13.72
In the case of high performance concrete with a high very early compressive
strength, there is a possibility that, because of a limited amount of hydration, the bond
between the aggregate and the matrix is not commensurately developed.
Consequently, with high strength at very early ages, flexural strength and the modulus

740
of elasticity are likely to be lower than would be expected from the usual relations
between these properties and the compressive strength.13.99
The elastic deformation of high performance concrete is of particular interest.
Because the modulus of elasticity of the very strong hardened cement paste and of the
aggregate differ less from one another than in medium strength concrete, the
behaviour of high performance concrete is more monolithic and the strength of the
aggregate–matrix interface is higher. There is, therefore, less bond microcracking, and
the linear part of the stress–strain curve extends up to a stress which may be as high as
85 per cent of the failure stress, or even higher (see Fig. 13.9). Subsequent failure
takes place through the coarse aggregate particles as well as through the matrix. Thus,
the coarse aggregate particles do not act as crack arrestors so that failure is rapid.13.71

Fig. 13.9. Typical stress–strain curves for concretes of different strengths13.71

741
 A reasonable relation between the modulus of elasticity of concrete, Ec, and its 28-
day compressive strength, both in MPa, is:13.91

In psi units, this equation becomes:

It is doubtful that this expression is valid at strengths well above 83 MPa (12 000
psi): generally, the modulus of elasticity at very high strengths is lower than would be
expected from an extrapolation of the above expressions. Some Japanese data on the
modulus of elasticity of concrete with strengths between 75 and 140 MPa (11 000 and
20 000 psi) are shown in Fig. 13.10.13.81

Fig. 13.10. Relation between the modulus of elasticity and compressive strength
for high strength concretes with a water/cement ratio of 0.25 tested at different
ages (based on ref. 13.81)

742
 Because of the strong bond between the coarse aggregate and the matrix, the
elastic properties of aggregate have a considerable influence on the modulus of
elasticity of the concrete.13.73 In consequence, the relation between the modulus of
elasticity of high performance concrete and its strength is much less consistent than in
the case of usual concrete; this is true regardless of the particular relation
used.13.73 Accordingly, for structural design purposes the modulus of elasticity of high
performance concrete should not be assumed to be a simple function of compressive
strength.

743
Chapter 14. Selection of concrete mix proportions (mix design)

It can be said that the properties of concrete are studied primarily for the purpose of
selection of appropriate mix ingredients, and it is in this light that the various
properties of concrete will be considered in this chapter.
In the British usage, the selection of the mix ingredients and their proportions is
referred to as mix design. This term, although common, has the disadvantage of
implying that the selection is a part of the structural design process. This is not correct
because the structural design is concerned with the required performance of concrete,
and not with the detailed proportioning of materials that will ensure that performance.
The American term mixture proportioning is unexceptional, but it is not used on a
world-wide basis. For this reason, in this book the expression at the head of the
chapter, sometimes abbreviated to mix selection, will be adopted.
Although the structural design is not normally concerned with mix selection, the
design imposes two criteria for this selection: strength of concrete and its durability. It
is important to add an implied requirement to the effect that workability must be
appropriate for the placing conditions. The workability requirement applies not only
to, say, slump at the time of discharge from the mixer but also to a limitation on the
slump loss up to the time of placing of concrete. Because of the dependence of the
required workability upon the site conditions, workability should generally not be
fixed prior to the consideration of the construction procedure.
In addition, the selection of mix proportions has to take into account the method of
transporting the concrete, especially if pumping is envisaged. Other important criteria
are: setting time, extent of bleeding, and ease of finishing; these three are interlinked.
Considerable difficultues can arise if these criteria are not properly taken into account
during the selection of the mix proportions or when adjusting these proportions.
The selection of mix proportions is thus, simply, the process of choosing suitable
ingredients of concrete and determining their relative quantities with the object of
producing as economically as possible concrete of certain minimum properties,
notably strength, durability, and a required consistency.
Cost considerations
The preceding sentence stresses two points: that the concrete is to have certain
specified minimum properties, and that it is to be produced as economically as
possible – a common enough requirement in engineering.
The cost of concreting, as of any other type of construction activity, is made up of
the costs of the materials, plant, and labour. The variation in the cost of material arises
from the fact that cement is several times dearer than aggregate, so that, in selecting
the mix proportions, it is desirable to avoid a high cement content. The use of
comparatively lean mixes confers also considerable technical advantages, not only in
the case of mass concrete where the evolution of excessive heat of hydration may
cause cracking, but also in structural concrete where a rich mix may lead to high
shrinkage and cracking. It is, therefore, clear that to err on the side of rich mixes is not
desirable, even if the cost aspect is ignored. In this connection it should be
remembered that the different cementitious materials vary in cost per unit mass, being,
with the exception of silica fume, cheaper than Portland cement. Their influence on

744
the different properties of concrete also varies, as discussed in the appropriate
chapters.
In estimating the cost of concrete, it is essential to consider also the variability of
its strength because it is the ‘minimum’, or characteristic, strength (see p. 734) that is
specified by the designer of the structure, and is indeed the criterion of acceptance of
the concrete, while the actual cost of the concrete is related to the materials producing
a certain mean strength. This touches very closely on the problem of quality control. It
should be borne in mind that a higher level of quality control represents a higher
expenditure both on supervision and on batching equipment, and there are occasions
when careful mix selection and quality control may not be justified. The decision on
the extent of quality control, often an economic compromise, will thus depend on the
size and type of construction. It is essential that the degree of control is estimated at
the outset of the process of selection of mix proportions, so that the difference
between the mean and the minimum, or characteristic, strength is known.
The cost of labour is influenced by the workability of the mix: workability
inadequate for the available means of compaction results in a high cost of labour (or
in insufficiently compacted concrete). Dealing with blockages in pumping is also
labour-intensive. The exact cost of labour depends on the details of organization of
the job and the type of equipment used, but this is a specialized topic.
Specifications
This large topic cannot be dealt with in this book and will be considered only in so far
as the type of specification affects the mix selection.
In the past, specifications for concrete prescribed the proportions of cement, and
fine and coarse aggregate. Certain traditional mixes were thus produced but, because
of the variability of the mix ingredients, concretes having fixed cement–aggregate
proportions and a given workability vary widely in strength. For this reason, the
minimum compressive strength was later added to other requirements. When the
strength is specified, the prescription of proportions makes the specification unduly
restrictive where good quality materials are available, but elsewhere it may not be
possible to achieve an adequate strength using the prescribed mix proportions. This is
why, sometimes, clauses prescribing the grading of aggregate and the shape of the
particles were added to the other requirements. However, the distribution of
aggregates in many countries is such that these restrictions are often uneconomic. In
this connection it should be noted that, with the exception of specialized construction,
such as nuclear containment vessels, only locally available aggregates are used;
transportation over long distances is prohibitively expensive.
More generally, specifying at the same time strength as well as mix ingredients
and their proportions, and also the aggregate shape and grading, leaves no room for
economies in the mix selection, and makes progress in the production of economic
and satisfactory mixes on the basis of the knowledge of the properties of concrete
impossible.
It is not surprising, therefore, that the modern tendency is for specifications to be
less restrictive. They lay down limiting values but sometimes give also as a guide the
traditional mix proportions for the benefit of the contractor who does not wish to use a
high degree of quality control. The limiting values may cover a range of properties;
the more usual ones are:
1. ‘Minimum’ compressive strength necessary from structural considerations;

745
 2. Maximum water/cement ratio and/or minimum cement content and, in certain
conditions of exposure, a minimum content of entrained air to give adequate
durability;
3. Maximum cement content to avoid cracking due to the temperature cycle in
mass concrete;
4. Maximum cement content to avoid shrinkage cracking under conditions of
exposure to a low humidity; and
5. Minimum density for gravity dams and similar structures.
In addition, the nature of the cementitious materials, sometimes by a specific
requirement with respect to the type or composition of cement, at other times by
proscription, may be included in the specification.
All these various requirements must be satisfied in the selection and proportioning
of mix ingredients.
Specification of quantities almost invariably includes associated tolerances on the
various quantities. With respect to strength, most national specifications lay down
clear requirements. The tolerances on cement content and water/cement ratio are
generally less clear but equally important. Particularly critical is the tolerance on
cover to reinforcement which, albeit not a ‘mix-proportions item’, is closely linked to
the specified strength of concrete and to its cement content from the durability
standpoint. The tolerance on cover must be explicitly specified and should be
logically associated with the tolerance on strength or on the cement content.
The British approach, given in BS EN 206-1 : 2000 and the complementary BS
8500-2 : 2002, is to recognize four methods of specifying concrete mixes. A designed
mix is specified by the designer principally in terms of strength, cement content, and
water/cement ratio; compliance relies on strength testing. A prescribed mix is
specified by the designer in terms of the nature and proportions of mix ingredients;
the concrete producer simply makes the concrete ‘to order’. The assessment of mix
proportions is used for compliance purposes, strength testing not being routinely used.
The use of prescribed mixes is advantageous when particular properties of concrete,
for instance with respect to its finish or abrasion resistance, are required. However, a
prescribed mix should be specified only when there are sound reasons for assuming
that it will have the required workability, strength, and durability.
A standardized mix is based on ingredients and proportions fully listed in BS
5328-2 : 2002 for several values of compressive strength up to 25 MPa, measured on
cubes. The fourth and last type of mix is the designated mix, for which the concrete
producer selects the water/cement ratio and the minimum cement content, using a
table of structural applications coupled with standard mixes. This approach can be
used only if the concrete producer holds a special certificate of product conformity
based on product testing and surveillance, coupled with certification of quality
assurance.
Standard mixes are used only in minor construction such as housing. Designated
mixes, although they can be used for strengths up to 50 MPa, are limited in
application to routine construction. It is, therefore, only in the selection of designed
and prescribed mixes that a full knowledge of properties of concrete can be used.
These four types of mixes are varied somewhat in BS 8500-2 : 2002.
In the American practice, when there is no experience on the basis of which mix
proportions could be selected and trial mixes made, it is necessary to base the mix

746
proportions on standard proportions which, in order to be safe, are perforce very
stringent. This approach can be used only for low strength concrete. For example,
ACI 318-0214.8 prescribes, for a specified 28-day compressive strength (measured on
cylinders) of 27 MPa (4000 psi), a maximum water/cement ratio of 0.44 in the case of
non-air-entrained concrete, and 0.35 in airentrained concrete. In the latter case, higher
strengths require a proper use of trial mixes, but, in the case of non-air-entrained
concrete, ACI 318-9514.8 allows the use of a water/cement ratio of 0.38 for concrete
with a specified 28-day strength of 31 MPa (4500 psi).
The process of mix selection
The basic factors which have to be considered in determining the mix proportions are
represented schematically in Fig. 14.1. The sequence of decisions is also shown down
to the quantity of each ingredient per batch. There are, of course, variations in the
exact method of selecting the mix proportions. For instance, in the excellent method
of the American Concrete Institute14.5 (see p. 754), the water content in kilograms per
cubic metre (or pounds per cubic yard) of concrete is determined direct from the
workability of the mix (given the maximum size of aggregate) instead of being found
indirectly from the water/cement ratio and the cement content.

Fig. 14.1. Basic factors in the process of mix selection

It should be explained that an exact determination of mix proportions by means of
tables or computer data is generally not possible: the materials used are essentially
variable and many of their properties cannot be assessed truly quantitatively. For
example, aggregate grading, shape and texture cannot be defined in a fully
satisfactory manner. In consequence, all that is possible is to make an intelligent guess
at the optimum combinations of the ingredients on the basis of the relationships
established in the earlier chapters. It is not surprising, therefore, that in order to obtain
a satisfactory mix, we not only have to calculate or estimate the proportions of the

747
available materials but must also make trial mixes. The properties of these mixes are
checked and adjustments in the mix proportions are made; further trial mixes are
made in the laboratory until a fully satisfactory mix is obtained.
However, a laboratory trial mix does not provide the final answer even when the
moisture condition of aggregate is taken into account. Only a mix made and used on
the site can guarantee that all the properties of the concrete are satisfactory in every
detail for the particular job in hand. To justify this statement three points may be
mentioned. Firstly, the mixer used in the laboratory is generally different in type and
performance from that employed on site. Secondly, the pumping properties of the mix
may need to be verified. Thirdly, the wall effect (arising from the surface to volume
ratio) in laboratory test specimens is larger than in the full-size structure, so that the
fine aggregate content of the mix as determined in the laboratory may be
unnecessarily high.
It can be seen then that mix selection requires both a knowledge of the properties
of concrete and experimental data or experience.
Other factors, such as effects of handling, transporting, delay in placing, and small
variations in weather conditions may also influence the properties of concrete on the
site but these are generally secondary and necessitate no more than minor adjustments
in the mix proportions during the progress of work.
This may be an appropriate place to note that the mix proportions, once chosen,
cannot be expected to remain entirely immutable because the properties of the
ingredients may vary from time to time. In particular, it is difficult to know the
precise amount of free water in the mix because of the variation in the moisture
content of the aggregate, expecially the fine aggregate. The problem is even greater
with lightweight aggregate, especially in pumped concrete. Other variations occur in
the grading of aggregate, particularly its dust content, and in the temperature of the
concrete due to exposure of the ingredients and of the mixer to the sun or due to the
cement being hot. In consequence, periodic adjustments to the mix proportions are
necessary.
Mean strength and ‘minimum’ strength
Compressive strength is one of the two most important properties of concrete, the
other one being durability. Strength is of importance both per se and also in so far as
it influences many other desirable properties of hardened concrete. Basically,
the mean compressive strength required at a specified age, usually 28 days,
determines the nominal water/cement ratio of the mix. Figure 14.2 gives this relation
for concretes made in the late 1970s with British ordinary Portland cements cured at
normal temperatures. This figure is intended to serve as no more than an illustration
and, in any case, the strength values in the figure err on the safe side. If, however, one
batch of cement is to be used throughout the job, it is possible to take advantage of the
actual strength of the given cement, that is, to use an experimental relation between
strength and the water/cement ratio.

748
Fig. 14.2. Relation between compressive strength and water/cement ratio for 102
mm (4 in.) cubes of fully compacted concrete for mixes of various proportions
made with typical British ordinary Portland cements of the late 1970s. The
values used are conservative estimates
If curves of the type shown in Fig. 14.2 are used, the type of cement must be
known because the rate of hardening of cements of different types varies; when
different cementitious materials are used, the variation in the rate of gain of strength
can be even larger. However, beyond the age of one or two years the strengths of
concretes made with different cements tend to be approximately the same.
Structural design is based on the assumption of a certain minimum strength of
concrete, but the actual strength of concrete produced, whether on site or in the
laboratory, is a variable quantity (see p. 639). In selecting a concrete mix we must,
therefore, aim at a mean strength higher than the minimum.
The distribution of strength of test specimens can be described by the mean and the
standard deviation. As mentioned on p. 641, the distribution of strength of concrete
test specimens is assumed to be normal (Gaussian). For practical purposes, such an
assumption is acceptable, even though examples of skewness have been reported: in
low strength concrete by McNicoll and Wong,14.23 and in high strength concrete by
Cook14.24 and also in ACI 363R-92.14.12 The assumption of normal distribution errs on

749
the safe side with respect to the number of test results expected to fall below the
specified value of strength.14.25
From the knowledge of the probability of a specimen having a strength differing
from the mean by a given amount (Table 14.1) we can define the ‘minimum’ strength
of a given mix. No absolute minimum can be specified because, from the statistical
viewpoint, there is always a certain probability of a test result falling below a
minimum, however low it is set; to make this probability extremely low would be
uneconomical. It is, therefore, usual to define the ‘minimum’ as a value to be
exceeded by a predetermined proportion of all test results, usually 95 per cent when
single test results are considered, and 99 per cent when a running average of three or
four test results is used.

Table 14.1. Percentage of Specimens Having a Strength Lower than (Mean –k ×

Standard Deviation)

The approach of the Building Code of the American Concrete Institute, ACI 318-
02 is based, in essence, on two requirements for the ‘minimum’ strength,
14.8
in
relation to the mean strength . First, there is a required probability of 1 per cent
that the average of three consecutive tests (a test being the average of two cylinders)
is smaller than the design strength. Second, there is a required probability of 1 per
cent that an individual test result falls below the design strength by more than 3.5 M
Pa (500 psi). In terms of standard deviation, σ, the first of these can be written as:

and the second (in MPa units) as

The two conditions are equivalent when the standard deviation σ is approximately 3.5
MPa (500 psi). When it is larger, the first condition is the more severe of the two.
We should note that no absolute limit value is laid down: the approach is
probabilistic so that failure to meet these requirements once in a 100 times is inherent
in the system. Such failure should not be a sufficient reason for rejection of the
concrete. It can be added that all specification schemes imply a risk of wrong
rejection and of wrong acceptance: it is the two risks that have to be judiciously
balanced.14.31

750
 The value of the standard deviation to be used in the expression of ACI 318-
02 given above is the value obtained experimentally in previous construction under
14.8

similar conditions, using similar materials, to produce concrete of similar strength. In

the absence of such an experimental value of the standard deviation, ACI 318-
0214.8 prescribes margins by which the mean compressive strength has to exceed the
specified value of strength. These margins are very substantial, ranging from 7 MPa
(1000 psi) when the specified strength is less than 21 MPa (3000 psi), to 10 MPa
(1400 psi) when the specified strength is above 35 MPa (5000 psi).
According to ACI 318-0214.8 and ASTM C 94-09a, compliance with the specified
value of strength, is achieved when both of the following requirements have been
satisfied:
(a) The average value of all sets of three consecutive test results is at least equal
to ; and
(b) No test result falls below by more than 3.5 MPa (500 psi).
It should be recalled that a test result is the average value of the strengths of two
test cylinders from the same batch of concrete, tested at the same age. The average
value of three consecutive test results is a running average; this means that a test
result number N appears in three sets as follows: N – 2, N – 1, N; N – 1, N, N + 1;
and N, N + 1, N + 2. Thus, if the value of test number N is very low, it can
significantly depress one or two or three average values. Consequently, all concrete
represented by tests numbered from N – 2 to N + 2 is deemed not to comply with the
specification. However, occasional failure to comply with the requirements of ACI
318-0214.8 has to be expected (probably once in 100 tests) so that automatic rejection of
the relevant concrete should not follow.
The requirements of Eurocode 2 : 2004 parallel those of ACI 318-08, referred to
earlier. A test result is the average of the strengths of two specimens but, in the British
practice, cubes are used. The British approach is to use a characteristic strength,
defined as the value of strength below which 5 per cent of all possible test results are
expected to fall; the margin between the characteristic strength and the mean strength
is chosen to achieve this probability. Compliance with the specified value of strength
is achieved when both of the following requirements have been satisfied:
(a) The average value of any four consecutive test results exceeds the specified
characteristic strength by 3 MPa (450 psi); and
(b) No test result falls below the specified characteristic strength by more than 3
MPa (450 psi).
Similar requirements are prescribed for flexural tests: the values in (a) and (b) in
the preceding paragraph are then 0.3 MPa (45 psi).
The topic of compliance cannot be treated adequately in this book but some
statements are worth making. It is not possible to discriminate absolutely between
satisfactory and unsatisfactory concrete, short of testing all of it! The object of testing
is to discriminate adequately so as to achieve a balance between the producer’s risk of
‘good’ concrete being rejected and the consumer’s risk of ‘bad’ concrete being
accepted. The balance is governed by the extent of testing as well as by the rules
used.14.31
Variability of strength

751
It may be remembered (p. 640) that the abscissa of any point on the normal
distribution curve is expressed in terms of the standard deviation σ, and the number of
specimens whose strength differs from the mean by more than kσ is represented by
the appropriate proportional area under the normal curve and is given in statistical
tables (Table 14.1).
Thus, if the mean strength of a sample of test specimens is , and the percentage
of specimens whose strength may fall below a certain value ( – kσ) is specified, then
the value of k can be found from statistical tables, and the actual difference between
the mean and the minimum, k σ, will depend only on the value of the standard
deviation σ. This is illustrated in Fig. 14.3. Because the cement content of the mix of a
given workability is related to the mean strength, it can be seen that the larger the
standard deviation the higher the cement content required for a given minimum
strength.

Fig. 14.3. Normal distribution curves for concretes with a minimum strength
(exceeded by 99 per cent of results) of 3000 psi (20.6 MPa):

The difference ( – k σ) can also be expressed in terms of the coefficient of

variation, C = σ/ , as (1 – kC). The two methods of estimating the minimum
strength are identical when applied to concrete of the same mean strength but, when
the data obtained for one mix are used to predict the variability of a mix of different
strength, the result will depend on whether the standard deviation or the coefficient of
variation is unaffected by the change in strength.

752
 If a constant standard deviation is assumed then, knowing the estimated value of
the standard deviation σ for one mix, we can calculate the mean strength of any other
mix by adding a constant value kσ to the minimum. This difference between the mean
and the minimum would be constant for the same process of manufacture of concrete.
On the other hand, if the coefficient of variation is assumed to be constant, the
minimum strength would form a fixed proportion of the mean. These two situations
are illustrated by the following numerical example.
Let us assume that concrete produced and tested under a given set of conditions
has a mean strength of 25 MPa (3600 psi) with a standard deviation of 4 MPa (580
psi). According to ACI 214-77 (Reapproved 1989),14.18 this represents ‘good’ control
(see Table 14.2). The coefficient of variation is (4/25) × 100, that is 16 per cent. For
illustration purposes, let us assume that the required ‘minimum’ strength is defined as
the strength exceeded by 99 per cent of all results. Using Table 14.1, we find that this
‘minimum’ strength is:

25 – 2.33 × 4 = 15.7 MPa.

Table 14.2. Classification of Standard of Control of Concretes with Strengths up

to 35 MPa (5000 psi) According to ACI 214-77 (Reapproved 1989)14.18

Imagine now that it is desired to produce, under the same conditions and using the
same materials, a concrete with a ‘minimum’ strength of, say, 50 MPa. The mean
strength aimed at, according to the ‘coefficient of variation method’, would be:

where as the figure given by the ‘standard deviation method’ would be:

50 + 2.33 × 4 = 59 MPa.

The practical significance of the difference between the two methods is clearly
reflected in the cost of producing a 79 MPa concrete as compared with a 59 MPa
concrete under the same control.
An estimate of the difference between the mean strength and the specified
‘minimum’ or characteristic strength, must be made at the outset of the process of mix
selection. The advice of ACI 214-77 (Reapproved 1989)14.18 is non-committal: “The
decision as to whether the standard deviation or the coefficient of variation is the
appropriate measure of dispersion to use in any given situation depends on which of
the two measures is the more nearly constant over the range of strengths characteristic

753
of the particular situation.” Nevertheless, ACI 214-77 (Reapproved 1989)14.18 includes
a table, reproduced here as Table 14.2, based on the assumption of a constant standard
deviation, for concretes with strengths up to 35 MPa (5000 psi). However, discussion
in Committee 214 of ACI continues as opinions are divided. It should be pointed out
that the convenience of calculations and simplicity of approach, often brought into the
discussions, are not the correct criteria upon which to decide whether it is the standard
deviation or the coefficient of variation that should be used. What matters is the actual
behaviour of concrete in construction.
The recommendations of ACI 214-77 (Reapproved 1989)14.18 are based on concretes
used up to the mid-1970s, and such concretes did not often have a cylinder strength in
excess of 35 MPa (5000 psi). It is, therefore, questionable whether the approach of
ACI 214-77 (Reapproved 1989) necessarily applies to high strength concrete with a
28-day compressive strength in excess of 80 MPa (12 000 psi), let alone in the region
of 120 MPa (17 000 psi).
Before discussing the variability of high strength concrete, it may be useful to
consider changes in concrete-making which occurred between, say, 1970 and the mid-
1990s. There is no doubt that the batching equipment has been greatly improved, with
the consequence of a much smaller variability in the mix proportions between batches.
As a result, the between-test standard deviation of compressive strength test results
can be expected to be smaller than in the past. On the other hand, there are few
grounds for expecting the within-test variation, which arises from the operator error
and the testing-machine error, to be different from what it was in the 1970s. Thus, it is
likely that the overall standard deviation of test results is smaller, but not much
smaller, than in the past.
In this connection it is useful to point out that the within-test and between-test
standard deviations are not arithmetically additive; it is the variances that are additive.
For instance, if the within-test standard deviation is 3 MPa, and the between-test
standard deviation is 4 MPa, then the overall standard deviation is (32 + 42)1/2 = 5 MPa.
A reduction in the between-test standard deviation to 3 MPa, while the within-test
variation remains unaltered, would reduce the overall standard deviation to (32 +
32)1/2 = 4.25 MPa. Thus, in this particular example, a reduction in the between-test
standard deviation of 1 MPa has reduced the overall standard deviation by only 0.75
MPa.
Returning now to high strength concrete, it is reasonable to assume that such
concrete is produced only in modern plants with low-variability batching and with
highly skilled and motivated personnel. However, the same plants also produce low-
or medium-strength concrete whose variability will also be lower than the variability
of concrete of similar strength produced in the 1970s. It follows that viewing the
variability of high strength concrete (which is all of recent production) against the
background of concretes of the 1970s gives a distorted picture.
The approach of ACI 363R-9214.22 is to recognize that the standard deviation of
“high strength concrete becomes uniform in the range of 500 to 700 psi (3.5 to 4.8
MPa)”. Thus, the coefficient of variation decreases with an increase in strength and, in
the words of ACI 363R-92, “the standard deviation method of evaluation appears to
be a logical quality control procedure”.
The problem of the constant standard deviation or constant coefficient of variation
is still controversial but, for a constant degree of control, laboratory test data, as well
as some results of actual site tests, have been shown to support the suggestion of a

754
constant coefficient of variation for well-compacted concretes of different mix
proportions with strengths higher than about 10 MPa (1500 psi) (Fig. 14.4). On the
other hand, the median values of standard deviation for different characteristic
strengths measured in Swedish ready-mixed plants in 1975 suggest a constant
standard deviation. The actual values are as follows:14.32

Fig. 14.4. Relation between the standard deviation and mean strength for
laboratory test cubes; regression line shown14.26

The distribution of the standard deviation for all classes of concretes is shown in Fig.
14.5.

755
 Fig. 14.5. Distribution of the standard deviation (in 0.5 MPa intervals) for all
classes of concrete from ready-mixed plants in Sweden in 197514.32

756
 Swiss Standard SIA 162 (1989),14.21 presumably based on Swiss experience,
assumes that the standard deviation is independent of strength for strengths up to 45
MPa measured on single 200 mm cubes.
Surveys of test data on a large number of construction sites suggest that neither the
assumption of a constant standard deviation or of a constant coefficient of variation at
all ages is generally valid for site-made test specimens. From Newlon’s review of the
problem,14.30 it appears that the coefficient of variation is constant up to some limiting
value of strength but, for higher strengths, the standard deviation remains constant
(see Fig. 14.6). Different investigators have found different values of this limiting
strength, which may well depend on site conditions and general construction practice.

Fig. 14.6. Relation between the standard deviation and the mean strength of test
specimens obtained from surveys of site data14.30
It is possible, however, to suggest some generalizations. Figure 14.6 indicates that
for single cubes, the limiting strength is about 34 MPa (5000 psi); for averages of two
cylinders, it is about 17 MPa (2500 psi); and an international survey involving both
cubes and cylinders, tested singly and in pairs, gives an intermediate value of about 31
MPa (4500 psi). The factors accounting for these differences are not clear, but
probably the lower variability of cylinders compared with cubes (cf. p. 596) is
relevant. It may also be noted that, for the same intrinsic strength, the cylinder
strength is lower than the cube strength. All these data apply to tests at a fixed age: for
the same source of concrete, an increase in age leads to a reduction in the coefficient
of variation, but the standard deviation increases; thus, it is the strength level and not
merely the concrete-making that is relevant.
It is probable that neither the standard deviation nor the coefficient of variation is
constant over a wide range of strengths of concrete made in the same plant. This view
is supported by the Commentary on ACI 318-0214.8 which states that “there may be an
increase in standard deviation when the average strength level is raised by a

757
significant amount, although the increment of increase in standard deviation should be
somewhat less than directly proportional to the strength increase”. The British
approach14.11 is to assume the standard deviation to be proportional to strength up to a
value of 20 MPa (2900 psi) but for higher strengths, that is, for structural concrete, the
standard deviation is assumed to be constant. In practice, therefore, it is best to
establish experimental relations between the mean and minimum strengths under
actual site conditions.
With respect to the variability of flexural strength, Greer14.2 and Lane14.3 confirmed
earlier findings to the effect that both the within-test standard deviation and the
between-test standard deviation are independent of the value of the flexural strength.
A typical value of between-test standard deviation, when the level of control is good,
is below 0.4 MPa (60 psi) (see Fig. 14.7).

Fig. 14.7. Relation between standard deviation and 28-day flexural strength
determined in pavement construction14.3
Quality control
It is apparent from Fig. 14.3 that the lower the difference between the minimum
strength and the mean strength of the mix the lower the cement content that need be
used. The factor controlling this difference for concrete of a given level of strength is
the quality control. By this is meant the control of variation in the properties of the
mix ingredients and also control of accuracy of all those operations which affect the
strength or consistency of concrete: batching, mixing, transporting, placing, curing,

758
and testing. Thus, quality control is a production tool, and one of its reflections is the
standard deviation.
The variation in the strength of cement was discussed in Chapter 7. On a large
project, it is possible to eliminate most of this variation by obtaining cement from one
source only, when advantage can be taken of the actual strength of the cement to be
used.
The influence of the variation in the grading of aggregate was stressed in Chapter 3,
and this factor is particularly important when the mix is controlled by workability
requirements: for the workability to be kept constant, a change in grading may require
an increase in water content with a consequent drop in strength.
Variations in strength of concrete arise also from inadequate mixing, insufficient
compaction, irregular curing, and variations in testing procedures – all discussed in
the appropriate chapters. The need for control of these factors on the site is obvious.
Changes in the moisture content of aggregate, unless carefully compensated for by
the amount of added water, also seriously affect the strength of concrete. To minimize
these changes, stockpiles should be arranged so that the aggregate is allowed to drain
before use; also, the mixer operator should be well trained in maintaining a constant
workability of the mix.
A standard deviation can be ascribed to each factor separately, although in some
cases the magnitude of the individual effects cannot be determined. As already
mentioned the various standard deviations are additive in the root-square-form, so that
if σ1 and σ2 are ascribed to two causes the resultant standard deviation
is . This is important to remember as the assumption of arithmetic
addition would lead to a gross over-estimate of the total standard deviation. The
knowledge of individual contributions of various factors to the overall variation,
obtained by statistical methods, is of value in deciding whether taking some measures
to reduce variation is economic, or whether the reduction in variability is
disproportionately small for the cost of improved control.
Quality control is sometimes taken to be synonymous with production of high-
strength concrete. This is certainly not so, as low-strength concrete can be
manufactured under good control, and this is indeed practised in the case of
construction of massive structures where obtaining large quantities of lean concrete of
low variability results in large savings. The degree of control is evaluated by the
variation in test results; various statistical techniques are available.
For the sake of completeness, quality assurance should be mentioned. This is an
administrative control system “implemented through quality assurance programmes to
provide a means of controlling quality-affecting activities to predetermined
requirements”.14.7 Thus, quality assurance is a management tool comforting to the
owner of the structure, but quality assurance per se does not produce concrete
appropriate for the given conditions.
Factors governing the selection of mix proportions
It may be convenient at this stage to restate the basic objective: we are to determine
the proportions of the most economical concrete mix that will be satisfactory both in
the fresh and in the hardened state. To achieve this, we shall now consider the various
factors of Fig. 14.1 and follow the sequence of decisions right down to the final

759
choice of mix proportions. It may be recalled that water/cement ratio and strength
have already been discussed.
Durability
It has been stated, on more than one occasion, that the selection of mix proportions
must satisfy not only the strength requirements but must also ensure adequate
durability. However, there does not yet exist a generally agreed and reliable approach
to the selection of mix of proportions required for durability under any given
conditions. One reason for this situation is the very wide range of exposure
circumstances, including the extremely onerous conditions in very hot and notionally
arid coastal areas. In these areas, protection of reinforcing steel from corrosion
strongly affects the selection of mix proportions of the concrete in the cover zone.
The nowadays widely accepted recognition of a specific durability requirement in
the mix selection is in contrast to the previous belief that reinforced concrete was
inherently durable and would remain in service for a long time without repair: the
maxim was ‘strong concrete is durable concrete’. For instance, British Standard Code
of Practice CP 114 (1948)14.12 stated: “No structural maintenance should be necessary
for dense concrete constructed in accordance with this code.” Even the 1969 edition
of the same code of practice14.10 limited itself to the statement: “The greater the severity
of the exposure the higher the quality of concrete required . . .”
The factors influencing durability were discussed in Chapters 10 and 11; simple
means of specifying the mix proportions so as to achieve the required durability will
now be considered. The word ‘simple’ is used in recognition of the fact that the
penetrability of concrete, which plays a crucial role in its durability, cannot be directly
controlled in the production of concrete. Hence, reliance is necessary on the
water/cement ratio, cement content, compressive strength – indeed, any one of these,
or two, or all three at the same time, can be used. It is worth reiterating that, whatever
mix proportions are chosen, the concrete must be capable of full compaction using the
means available, and that such compaction must be achieved in practice.
The American Concrete Institute Building Code 318-0214.8 devotes a separate
chapter to durability requirements. With respect to exposure to freezing and thawing,
ACI 318-02 requires, for normal weight concrete, a specified maximum water/cement
ratio, and for lightweight aggregate concrete, a specified minimum strength; these are
shown in Table 14.3. The reason for this difference in approach in the two types of
concrete is that it is not practicable to control the water/cement ratio of the lightweight
aggregate concrete. In addition, all concretes require air entrainment, the total air
content being specified according to the conditions of exposure and the maximum size
of aggregate used (see Table 11.3). The limitations on the amounts of fly ash and
ground granulated blastfurnace slag when de-icing agents are used, prescribed by ACI
318-0214.8 are given on p. 668.

Table 14.3. Requirements of ACI 318-0214.8 for Concrete Exposed to Freezing and
Thawing

760
 The requirements suggested in the U.S. Strategic Highway Research
Program,14.14 are more stringent than those of ACI 318-02: the water/cement ratio is not
to exceed 0.35 so as to ensure discontinuous capillaries in the cement paste after one
day’s curing.
British Standard BS 5328-1 : 1997 contains an elaborate classification of exposure
and recommends appropriate values of the maximum water/cement ratio, minimum
cement content, and 28-day compressive strength. These recommended values are
likely to be inadequate in climates other than temperate, and even under British
conditions they may be somewhat optimistic. One of these recommendations is to the
effect that a maximum water/cement ratio of 0.55, a minimum cement content of 325
kg/m3, and a 28-day characteristic strength of 40 MPa (measured on cubes) should be
used for concrete occasionally exposed to sea-water spray or de-icing agents or to
severe freezing conditions while wet. These recommendations are not endorsed in the
present book. All parts of BS 5328 have been withdrawn and replaced by BS EN 206-
1 : 2000 and BS 8500 : Parts 1 & 2 : 2002.
According to BS 5328-l : 1997 (withdrawn), a satisfactory strength “will generally
ensure that the limits on free water/cement ratio and cement content will be met
without further checking”. In view of the wide range of cements available world-wide,
this assumption may not be valid and it is not recommended in this book. In particular,
some cementitious materials increase the compressive strength of concrete but the
higher strength does not necessarily contribute to the resistance to freezing and
thawing or to carbonation.14.9 It is very much doubted that strength alone can be used
as an indicator of durability.
With respect to sulfate attack, BS 5328-1 : 1997 recommends both a maximum
value of the water/cement ratio and a minimum cement content, and also specifies the
type of cement to be used for various concentrations of sulfates in groundwater or in
soil. It is arguable that there is some inconsistency between the approach of the same
British Standard to the requirements for resistance to sulfate attack and to other

761
conditions of exposure for which strength alone can be used as a measure of
compliance. This situation may well be the consequence of the combination of our
inadequate understanding of the behaviour of concrete under various forms of attack,
coupled with practical difficulties of control of all aspects of mix ingredients and of
their proportions.
For resistance to sulfate attack, ACI 225R-9114.17 prescribes a maximum
water/cement ratio between 0.45 and 0.50 for the categories of exposure given
in Table 10.7. The cementitious materials to be used are also prescribed.
The cement content as such does not control durability: it does so only in so far as
it influences the water/cement ratio, which, in turn, influences strength. Moreover,
considering reliance on a minimum cement content, it should be remembered that,
while it is expressed in kilograms per cubic metre of concrete, durability depends
largely on the properties of hydrated cement paste. Thus, it is the cement content of
the paste that is relevant, and the volume of the cement paste (in a unit volume of
concrete) is smaller the larger the maximum size of aggregate. For this reason, BS
5328-1 : 1997 recommends an adjustment of the cement content as a function of the
maximum size of aggregate in the following manner. The specified cement content of
a mix with a maximum size of aggregate of 20 mm should be increased by 20
kg/m3 when the maximum size of aggregate is 14 mm, and by 40 kg/m3 when it is 10
mm. Conversely, when the maximum size of aggregate is 40 mm, the cement content
can be reduced by 30 kg/m3 as compared with concrete containing 20 mm aggregate.
It may be observed that, in the French approach, the cement content is assumed to be
inversely proportional to the fifth root of the maximum aggregate size; this ascribes to
the maximum size of aggregate a large influence on the required cement content.
If durability requires a certain maximum water/cement ratio, but structural
requirements are for a value of strength which can be readily achieved at a higher
water/cement ratio, a set of incompatible values of strength and water/cement ratio
should not be specified. Rather, a higher specified strength should be used so as to
correspond to the water/cement ratio required for reasons of durability. In this manner,
there will be no temptation on the part of the concrete producer to disregard the
water/cement ratio and to rely solely on an adequate level of strength.14.8 This higher
strength should be established prior to the commencement of the structural design so
that advantage can be taken of the use of a higher strength of concrete in the structural
design.
It has to be stated that little is known about the variability of the water/cement ratio
in in situ concrete. According to Gaynor,14.13 on well-controlled jobs the standard
deviation of the water/cement ratio is between 0.02 and 0.03. This high variability
may be a reflection of the fact that the total amount of free water in a given batch is
not easily ascertained. One reason for this is that, even if the moisture content of
aggregate is measured accurately, the result may not be representative of the given
batch.
The water/cement ratio alone does not determine the resistance of concrete to
chloride penetration: the type of cementitious material used greatly affects the
penetrability of the resulting concrete. In particular, concretes containing both ground
granulated blastfurnace slag and silica fume offer particularly good resistance.14.1 This
situation exemplifies the difficulty of basing the specification for durability on
strength alone. The same argument applies to the use of cement content alone.

762
 The nature of the cementitious materials to be used is of vital importance also
under other conditions of exposure. When concrete is to be subjected to chemical
attack, a suitable type of cement must be used but, if resistance to freezing and
thawing is the only durability requirement, the choice of the type of cement is
governed by other considerations, for instance, the development of early strength or of
a high heat of hydration for concreting in cold weather. Indeed, the beneficial
properties of the various cementitious materials, discussed in Chapter 13, should be
exploited in the selection of the cement. However, the limits on the maximum content
of fly ash and ground granulated blastfurnace slag, imposed by ACI 318-02)14.8 for
concrete exposed to de-icing agents, should be remembered (see p. 668).
Because the type of cement affects the early development of strength, it may be
necessary, with some cements, to use a low water/cement ratio to ensure a satisfactory
strength at early ages. Thus, strength, type of cement, and durability determine
between them the water/cement ratio required – one of the essential quantities in the
calculation of mix proportions.
Workability
So far, we have considered the requirements for the concrete to be satisfactory in the
hardened state but, as said before, properties when being transported, possibly
pumped, and placed are equally important. One essential at this stage is a satisfactory
workability. Selection of mix proportions which do not permit the achievement of
appropriate workability totally defeats the purpose of rational mix proportioning.
The workability that is considered desirable depends on two factors. The first of
these is the minimum size of the section to be concreted and the amount and spacing
of reinforcement; the second is the method of compaction to be used.
It is clear that when the section is narrow and complicated, or when there are
numerous corners or inaccessible parts, the concrete must have a high workability so
that full compaction can be achieved with a reasonable amount of effort. The same
applies when embedded steel sections or fixtures are present, or when the amount and
spacing of reinforcement make placing and compaction difficult. Because these
features of the structure are determined during its design, the necessary workability
must be ensured in the selection of mix proportions. On the other hand, when no such
limitations are present, workability may be chosen within fairly wide limits, but the
means of transportation and compaction must be decided upon accordingly; it is
important that the prescribed method of compaction is used during the entire progress
of construction. Advice on the appropriate value of slump and of means of
compaction for various types of construction is given in BS 5328-1 : 1997.
A property closely related to workability is cohesiveness. This depends largely on
the proportion of fine particles in the mix and, especially in lean mixes, attention must
be paid to the grading of the aggregate at the fine end of the scale. It is sometimes
necessary to make several trial mixes with different proportions of fine to coarse
aggregate in order to find the mix with an adequate cohesiveness.
While every mix should be cohesive so that uniform and well-compacted concrete
can be obtained, the exact importance of cohesiveness varies. For instance, where
concrete has to be hauled without agitation over a long distance or is handled down a
chute, or has to pass through reinforcement, possibly to some inaccessible corner, it is
essential that the mix be truly cohesive. In cases when the conditions leading to
segregation are less likely to be encountered, cohesion is of smaller importance, but a
mix which segregates easily must never be used.

763
Maximum size of aggregate
In reinforced concrete, the maximum size of aggregate which can be used is governed
by the width of the section and the spacing of the reinforcement. With this proviso, it
used to be considered desirable to use as large a maximum size of aggregate as
possible. However, it now seems that the improvement in the properties of concrete
with an increase in the size of aggregate does not extend beyond about 40 mm (1 in.)
so that the use of larger sizes may not be advantageous (see p. 174). In particular, in

high performance concrete, the use of aggregate larger than 10 to 15 mm ( to in.)

is counter-productive (see p. 678).
Furthermore, the use of a larger maximum size means that a greater number of
stockpiles has to be maintained and the batching operations become correspondingly
more complicated. This may be uneconomical on small sites but, where large
quantities of concrete are to be placed, the extra handling cost may be offset by a
reduction in the cement content of the mix.
The choice of the maximum size may also be governed by the availability of
materials and by their cost. For instance, when various sizes are screened from a pit, it
is generally preferable not to reject the largest size, provided this is acceptable on
technical grounds.
Grading and type of aggregate
Most of the remarks in the preceding section apply equally to the considerations of
aggregate grading because it is often more economical to use the material available
locally, even though it requires a richer mix (but provided it will produce concrete
free from segregation) rather than to bring in a better graded aggregate from farther
afield.
It has been stressed repeatedly that, although there are certain desiderata for a good
grading curve, no ideal gradings exist, and excellent concrete can be made with a
wide range of aggregate gradings.
The grading influences the mix proportions for a desired workability and the
water/cement ratio: the coarser the grading the leaner the mix which can be used, but
this is true within certain limits only because a very lean mix will not be cohesive
without a sufficient amount of fine material.
It is possible, however, to reverse the direction of choice: if the cement content is
fixed (e.g. a lean mix may be essential for massive concrete construction) then a
grading must be chosen such that concrete of given water/cement/aggregate
proportions and having a satisfactory workability can be made. Clearly, there are
limits on grading outside which it is not possible to make good concrete.
The influence of the type of aggregate should also be considered because its
surface texture, shape and allied properties influence the aggregate/cement ratio for a
desired workability and a given water/cement ratio. In selecting a mix, it is essential,
therefore, to know at the outset what type of aggregate is available.
An important feature of satisfactory aggregate is the uniformity of its grading. In
the case of coarse aggregate, this is achieved comparatively easily by the use of
separate stockpiles for each size fraction. However, considerable care is required in
maintaining the uniformity of grading of fine aggregate, and this is especially
important when the water content of the mix is controlled by the mixer operator on
the basis of a constant workability: a sudden change toward finer grading requires

764
additional water for the workability to be preserved, and this means a lower strength
of the batch concerned. Also, an excess of fine aggregate may make full compaction
impossible and thus lead to a drop in strength.
Thus, while narrow specification limits for aggregate grading may be unduly
restrictive, it is essential that the grading of aggregate varies from batch to batch
within prescribed limits only.
Cement content
All the factors considered up to now, including water/cement ratio, will determine
between them the aggregage/cement ratio or the cement content of the mix. To obtain
a clear picture of the various influences, Fig. 14.1 should once again be consulted.
The choice of the cement content is made either on the basis of experience or
alternatively from charts and tables prepared from comprehensive laboratory tests.
Such tables are no more than a guide to the mix proportions required because they
apply fully only to the actual aggregates used in their derivation. Moreover,
recommended proportions are usually based on aggregate gradings which have been
found to be satisfactory. When a significant departure from such gradings is necessary,
it may be useful to bear in mind some of the guidance ‘rules’ established as far back
as 1950. One of these ‘rules’ is: when there is an excess of particles smaller than
600 μm (No. 30 ASTM) sieve, the quantity of material passing the 4.76 mm ( in.)
sieve should be reduced by an amount up to 10 per cent of the total aggregate. On the
other hand, when there is an excess of particles in the 1.20 to 4.76 mm (No. 16 ASTM
to in.) size range, the quantity of fine aggregate should be increased. However, fine
aggregate with a large excess of particles between 1.20 mm (No. 16 ASTM) and 4.76
mm ( in.) sieves produces a harsh mix and may require a higher cement content for
a satisfactory workability.
In comparing various mixes, it is sometimes convenient to convert rapidly the
aggregate/cement ratio into the cement content or vice versa: Fig. 14.8 makes such a
conversion very easy.

765
 Fig. 14.8. Conversion chart for aggregate/cement ratio and cement content
(courtesy of Cement and Concrete Association)
Mix proportions and quantities per batch
With the water/cement ratio and the cement content known, there is no difficulty in
determining the proportions of cement, water, and aggregate. In practice, the
aggregate is supplied from at least two stockpiles, and the quantities of aggregate of
each size have to be given separately. This presents no problem because, in finding a
suitable grading, we already had to calculate the proportions of the different size
fractions of aggregate. The details of calculation are given in the example on p. 752.
For practical purposes, the mix quantities are given in kilograms or pounds per
batch. When cement is supplied in bulk, we choose the batch quantities so that their
sum is equal to the capacity of the mixer. When cement is supplied in bags, and there
is no provision for weighing it, it is preferable to choose the batch quantities so that

766
the mass of cement per batch is one bag or its multiple. The mass of cement is then
known accurately. In exceptional cases, a half-bag can be used, but other fractions
cannot be reliably determined and should never be used. Bag sizes are given on p. 7.
If a concrete mix of certain proportions is to be modified by the use of an
admixture, some changes in the quantity of some of the ingredients are necessary. An
important principle is to maintain the volume of the coarse aggregate in a unit volume
of concrete, and to adjust only the volume of fine aggregate. This is done by changing
the quantity of fine aggregate on an absolute volume basis by an amount equal and
opposite to the changes in the volume of water, entrained air and cement. The liquid
part of any admixture is considered to be a part of the mix water.
Calculation by absolute volume
The procedure so far described leads to the determination of values of the
water/cement ratio and the cement content or the aggregate/cement ratio, and also of
the relative proportions of the aggregates of various sizes, but does not give the
volume of fully compacted concrete produced by these materials. This volume is
obtained by a simple calculation, using the so-called absolute volume method, which
assumes that the volume of compacted concrete is equal to the sum of the absolute
volumes of all ingredients.
It is usual to calculate the quantities of ingredients to produce 1 cubic metre or 1
cubic yard of concrete. Then, if W, C, A1, and A2 are the required quantities by mass of
water, cement, fine aggregate, and coarse aggregate, respectively, we have, for the
cubic metre:

where ρ with the appropriate suffix respresents the specific gravity of each material.
In the Imperial or American system of units of measurement, since the density of
water (62.4) is expressed in pounds per cubic foot, the total volume of 1 cubic yard
has to be expressed as 27 cubic feet. For the cubic yard, therefore the corresponding
equation is:

The mix proportioning calculations give the values of W/C, C/(A1 + A2) and A1/A2,
whence the values of W, C, A1, and A2 can be found.
When an additional cementitious material, possibly with a specific gravity
different from that of Portland cement, is present, or when the coarse or fine aggregate
is in more than one stockpile, additional terms of similar form are added to the
equation. When entrained air is present, and its percentage is, say, a per cent of the
volume of concrete, the right-hand side of the ‘cubic yard’ equation would read:

For the ‘cubic metre’ equation, 27 is replaced by 1.

In the preceding equations, C represents the cement content in kilograms per cubic
metre or pounds per cubic yard of the concrete, and W is the water content in the same

767
units; the latter must not be confused with the water/cement ratio. In the United States,
the cement content used to be expressed in sacks of cement per cubic yard of concrete
and was referred to as cement factor; a sack weighs 94 lb.
If the aggregate contains free moisture whose mass is, say, m per cent of the mass
of the dry aggregate then the values of the mass of the added water W and of (wet)
aggregate must be adjusted. The mass of free water in A′ kg (lb) of aggregate is x such
that:

and the mass of dry aggregate is A = A′ – x. Hence, x = Am/100. This mass is added
to A to give the mass of wet aggregate per batch, A(1 + m/100), and is subtracted
from W to give the mass of added water, W – Am/100.
Generally, each size fraction of aggregate has a different moisture content, and the
correction should be applied to A1, A2, etc., with an appropriate value of m.
In the production of concrete of low strength, the determination of the moisture
content of aggregate can be dispensed with if the grading of aggregate is reasonably
constant and weigh-batching is used. Under those circumstances, a change in
workability caused by a variation in the moisture content of aggregate can be
prevented by an experienced mixer operator who can adjust the amount of added
water so that the workability, as judged by eye, remains constant. The water/cement
ratio remains then also sensibly constant. It should be stressed, however, that, if
concrete of specified proportions is to be consistently produced, then it is essential for
all ingredients, including moisture in the aggregate, to be precisely determined.
In the case of volume batching, no correction for moisture content need be made in
the case of coarse aggregate, but the bulking of fine aggregate must be allowed for
(see p. 134). The quantity of added water must be adjusted by the mixer operator as in
the case of weigh-batching.
Combining aggregates to obtain a type grading
While there are no ideal gradings – a point repeatedly made – it may be desirable or
required to proportion the available materials in such a way that the grading of the
combined aggregate is similar to a specific curve or lies between given limits. This
can be done by calculation or graphically. Both procedures are best illustrated by
means of examples.
In these examples, it is assumed that all aggregates have the same specific gravity.
The physical composition of concrete is, however, based on volumetric proportions. It
follows that, if the specific gravities of the different size fractions differ appreciably
from one another, the proportions required should be adjusted accordingly. This
approach is necessary in the calculation of mix proportions of lightweight aggregate
mixes when lightweight coarse aggregate and normal weight fine aggregate are used.
Suppose the gradings of the fine aggregate and the two coarse aggregate size
fractions are as listed in Table 14.4, and we are to combine the materials so as to
approximate to the coarsest grading of Fig. 3.15 (curve 1). On this curve, 24 per cent
of the total aggregate passes the 4.75 mm ( in.) sieve, and 50 per cent passes the
19.0 mm ( in.) sieve.

768
 Table 14.4. Example of Combining Aggregates to Obtain a Type Grading

Let x, y, z be the proportions of fine, 19.0 to 4.75 mm ( to in.), and 38.1 to 19.0
mm ( to in.) aggregates, respectively. Then, to satisfy the condition that 50 per
cent of the combined aggregate passes the 19.0 mm ( in.) sieve, we have:

1.0x + 0.99y + 0.13z = 0.5(x + y + z).

The condition that 24 per cent of the combined aggregate passes the 4.75 mm ( in.)
sieve can be written:

0.99x + 0.05y + 0.02z = 0.24(x + y + z).

From these two equations we find:

x:y:z = 1:0.94:2.59

that is, the three aggregates are combined in the proportions 1:0.94:2.59.
To find the grading of the combined aggregate we multiply columns (1), (2), and
(3) of Table 14.4 by 1, 0.94, and 2.59, respectively, the products being shown in
columns (4), (5), and (6). We now add these three columns (column 7) and divide the
sum by 1 + 0.94 + 2.59 = 4.53. The result, given in column (8), is the grading of the
combined aggregate. The grading is given to the nearest 1 per cent as, because of the
variability of the materials, any higher apparent accuracy has no meaning.
Figure 14.9 shows the grading of the combined aggregate, together with the type
curve which we are seeking to follow. Deviations are apparent, and indeed
unavoidable, because agreement with the type curve is generally possibly only at
specified points.

769
 Fig. 14.9. Grading of the aggregate for the example of Table 14.4
The graphical method is shown in Fig. 14.10. The two coarse aggregates are
combined first, using the percentage passing the 19.0 mm ( in.) sieve as a criterion.
Percentage passing is marked along three sides of a square. The values for the two
coarse aggregates are entered on two opposite sides, and the points corresponding to
the same sieve size are joined by straight lines. A vertical line is now drawn through
the point where the line joining the 19.0 mm ( in.) values intersects the horizontal
line representing the correct percentage of aggregate smaller than 19.0 mm ( in.). In
our case, (50 – 24) = 26 parts of aggregate coarser than 9.50 mm ( in.) sieve are to
pass the 19.0 mm ( in.) sieve while 50 parts are to be retained. The ratio is thus
26:(50 + 26), or 34 per cent of all coarse aggregate. A horizontal line is, therefore,
drawn through the 34 per cent point to intersect the 19.0 mm ( in.) line at A. A
vertical line through A gives the quantity of material 19.0 to 4.75 mm ( to in.) as a
percentage of the total coarse aggregate. In Fig. 14.10(a) this value is 24 per cent. The
vertical line gives also the grading of the combined coarse aggregate, and this is
combined with the fine aggregate in a similar manner to that already described (Fig.
14.10(b)). We find that 22 parts of fine aggregate are to be combined with 78 parts of
aggregate coarser than 4.75 mm ( in.) sieve. The aggregate is thus to be
proportioned as 22:(24/100) × 78:(76/100) × 78, or 1:0.85:1.69. The vertical line
through B (Fig. 14.10(b)) gives the combined grading of aggregate obtained by
proportioning the three aggregates in the ratio 1:0.85:2.69. This agrees with the
grading obtained earlier by calculation, but both methods are approximations based on
quantities passing only two specific sieve sizes.

770
 Fig. 14.10. Graphical method of combining aggregates (example of Table 14.4)

771
 It is possible to draw (in a figure of the type of Fig. 14.10(b)) envelopes of
standard gradings: because any vertical line represents a possible grading it is
immediately apparent whether or not a grading within the envelope can be obtained;
the range of proportions is then given by a point similar to B, corresponding to any
chosen vertical line.
American method of selection of mix proportions
The ACI Standard Practice ACI 211.1-9114.5 describes a method of selection of mix
proportions of concrete containing Portland cement alone or together with other
cementitious materials, and containing also admixtures. It should be emphasized that
the method provides a first approximation of mix proportions to be used in trial mixes.
In essence, the method of ACI 211.1-91 consists of a sequence of logical,
straightforward steps which take into account the characteristics of the materials to be
used. These steps will now be described.
Step 1: Choice of slump
At the time of mix proportioning, the slump will have been determined by the
exigencies of construction. It should be noted that slump should be specified not only
at the minimum end, but a maximum value should also be specified. This is necessary
to avoid segregation when the mix, which has not been selected to have a higher
slump, suddenly becomes ‘wet’.
Step 2: Choice of maximum size of aggregate
This, too, will have been decided, usually by the structural designer, bearing in mind
the geometric requirements of member size and spacing of reinforcement, or
alternatively for reasons of availability.
Step 3: Estimate of water content and air content
As discussed in Chapter 4, the water content required to produce a given slump
depends on several factors: the maximum size of aggregate, its shape, texture, and
grading; the content of entrained air; the use of admixtures with plasticizing or water-
reducing properties; and the temperature of concrete. Tables relating slump to these
properties have to be used, unless direct experience is available. One such table
is Table 4.1. Alternatively, the values recommended by ACI 211.1-9114.5 can be used; a
selection of these is given in Table 14.5. For use in practice, notes to this table and
comments in ACI 211.1-91, which are not reproduced here, should be taken into
account.

Table 14.5. Approximate Mixing Water and Air Content Requirements for
Different Slumps and Nominal Maximum Sizes of Aggregates given in ACI
211.1-9114.5

772
 The values of Table 14.5 are typical for well-shaped angular aggregates having
what is considered to be ‘good’ grading. When the coarse aggregate is rounded, the
water requirement per cubic metre of concrete can be expected to be reduced by about
18 kg in the case of non-air-entrained concrete, and by 15 kg for air-entrained
concrete. Water-reducing admixtures, and even more so superplasticizers, will
significantly reduce the values of water given in Table 14.5. It should be remembered
that the liquid part of admixtures constitutes a part of the mix water.
Table 14.5 also gives the values of the amount of entrapped air which can be
expected. These are useful in the calculation of density of compacted concrete and of
yield.
Step 4: Selection of water/cement ratio
There are two criteria for the selection of the water/cement ratio: strength and
durability. As far as the compressive strength is concerned, the average value aimed at
must exceed the specified ‘minimum’ strength by an appropriate margin (see p. 734).
The term ‘cement’ refers to the total mass of cementitious materials used; their choice
is governed by numerous factors, such as heat development, rate of gain of strength,
and resistance to various forms of attack, so that the type of blended cement to be
used has to be established at the outset of mix proportioning. It is for the actual
cement to be used that the relation between strength and the water/cement ratio has to
be established over a certain range of strengths.
As far as durability is concerned, the water/cement ratio may well be specified by
the structural designer or by an appropriate design code. What is vital is that the
water/cement ratio chosen is the lower of the two values emanating from strength and
durability requirements.

773
 When different cementitious materials are used, it should be remembered that they
have varying values of specific gravity: typical values are 3.15 for Portland cement,
2.90 for ground granulated blastfurnace slag, and 2.30 for fly ash.
Step 5: Calculation of cement content
The outcome of Steps 3 and 4 gives the cement content directly: it is the water content
divided by the water/cement ratio. If, however, from durability considerations, there is
a requirement for a certain minimum cement content, the larger of the two values
must be used.
Occasionally, from heat development considerations, the specification imposes a
maximum cement content. Clearly, this must be observed. Heat development is of
particular concern in mass concrete, and mix proportioning for that type of concrete is
specifically covered in ACI 211.1-91.14.5
Step 6: Estimate of coarse aggregate content
Here, the assumption is made that the optimum ratio of the bulk volume of coarse
aggregate to the total volume of concrete depends only on the maximum size of
aggregate and on the grading of fine aggregate. The shape of the coarse aggregate
particles does not directly enter the relation because, for instance, a crushed aggregate
has a greater bulk volume for the same mass (that is, a lower bulk density) than a
well-rounded aggregate. Thus, the shape factor is automatically taken into account in
the determination of the bulk density. Table 14.6 gives values of the optimum volume
of coarse aggregate when used with fine aggregates of different fineness moduli (see
p. 154). This volume is converted into mass of coarse aggregate per cubic metre of
concrete by multiplying the value from the table by the oven-dry rodded mass of the
aggregate (in kg/m3).

Table 14.6. Bulk Volume of Coarse Aggregate per Unit Volume of Concrete14.5

774
Step 7: Estimate of fine aggregate content
At this stage, the mass of fine aggregate is the only remaining unknown quantity. The
absolute volume of this mass can be obtained by subtracting the sum of the absolute
volumes of water, cement, entrained air, and coarse aggregate from the volume of
concrete, that is, 1 m3. For each ingredient, the absolute volume is equal to the mass
divided by the absolute density of the material (in kg/m3); the absolute density is the
specific gravity of the material multiplied by the density of water (1000 kg/m3).
The absolute volume of fine aggregate is converted into mass by multiplying this
volume by the specific gravity of the fine aggregate and by the density of water.
Alternatively, the mass of fine aggregate can be obtained directly by subtracting
the total mass of other ingredients from the mass of a unit volume of concrete, if this
can be estimated from experience. This approach is slightly less accurate than the
absolute volume method.
Step 8: Adjustments to mix proportions
As in any process of selection of mix proportions, trial mixes have to be made.
Advice, by way of some rules of thumb, for adjustments to the mix, is given in ACI
211.1-91.14.5 In general terms, it is important to remember that, if workability is to be
changed, but the strength is to remain unaffected, the water/cement ratio must remain
unaltered. Changes can be made in the aggregate/cement ratio or, if suitable
aggregates are available, in the grading of the aggregate; the influence of grading on
workability was discussed in Chapter 3.
Conversely, changes in strength but not in workability are made by varying the
water/cement ratio with the water content of the mix remaining unaltered. This means
that a change in the water/cement ratio must be accompanied by a change in the
aggregate/cement ratio so that the mass ratio

is approximately constant.
The above mix proportioning method of the American Concrete Institute can
readily be programmed for computer use; an example of manual calculation is given
later in this section.
Example
We require a mix with a mean 28-day compressive strength (measured on standard
cylinders) of 35 MPa and a slump of 50 mm, ordinary Portland cement being used.
The maximum size of well-shaped, angular aggregate is 20 mm, its bulk density is
1600 kg/m3, and its specific gravity is 2.64. The available fine aggregate has a
fineness modulus of 2.60 and a specific gravity of 2.58. No air entrainment is required.
For the sake of completeness, all steps, even when obvious, will be given.
Step 1: A slump of 50 mm is specified.
Step 2: The maximum size of aggregate of 20 mm is specified.
Step 3: From Table 14.5 for a slump of 50 mm and a maximum size of
aggregate of 20 mm (or 19 mm), the water requirement is approximately
190 kg per cubic metre of concrete.

775
 Step 4: From past experience, a water/cement ratio of 0.48 is expected to result
in concrete with a compressive strength, measured on cylinders, of 35
MPa. There are no special durability requirements.
Step 5: The cement content is 190/0.48 = 395 kg/m3.
Step 6: From Table 14.6, when used with a fine aggregate having a fineness
modulus of 2.60, the bulk volume of oven-dry rodded coarse aggregate
with a maximum size of 20 mm is 0.64. Given that the bulk density of
the coarse aggregate is 1600 kg/m3, the mass of coarse aggregate is 0.64
× 1600 = 1020 kg/m3.
Step 7: To calculate the mass of fine aggregate, we need first to calculate the
volume of all the other ingredients. The required values are as follows.

From the various steps, we can list the estimated mass of each of the ingredients in
kilograms per cubic metre of concrete as follows:

Mix selection for no-slump concrete

The ACI 211.1-9114.5 method of selection of mix proportions is intended for use with
concretes having a slump of at least 25 mm (1 in.). For no-slump concrete, some
modifications are required; these are given in ACI 211.3-75 (Revised 1987)
(Reapproved 1992).14.4
The primary modification is applied to the water requirement given in Table 14.5.
The values in that table for concrete with a slump of 75 to 100 mm (3 to 4 in.) are
taken as reference values. Assigning a relative value of 100 per cent to those reference
values of water requirement (in kg/m3), the water requirement at other values of
workability can be taken as a percentage given in Table 14.7. Three categories of no-
slump concrete are recognized: extremely dry, very stiff, and stiff. The same table
also gives relative values of water requirement at higher workabilities.

Table 14.7. Relative Mixing Water Requirements for Concretes with Different
Workabilities14.4

776
 The second modification to the procedure of ACI 211.1-91 for the purpose of mix
selection for no-slump concrete is in the values of the bulk volume of coarse
aggregate per unit volume of concrete: those given in Table 14.6 have to be multiplied
by factors listed in Table 14.8. Further details are given in ACI 211.1-91. Otherwise,
the procedure for no-slump concrete is similar to the procedure of 211.1-91, described
earlier.

Table 14.8. Factors to be Applied to the Volume of Coarse Aggregate Calculated

on the Basis of Table 14.6 for Mixes of Different Workabilities14.4

Mix selection for flowing concrete

Some special comments with respect to flowing concrete should now be made. First
of all, flowing concrete is described by ASTM C 1017-07 as concrete with a slump
greater than 190 mm ( in.) which has a cohesive nature. Commonly, flowing
concrete has a slump of 200 mm (8 in.) or a flow of 510 to 620 mm or a compacting
factor of 0.96 to 0.98. In the process of mix selection, it is convenient first to do so for
concrete with a slump of 75 mm (3 in.), the higher slump being then achieved by the
addition of a superplasticizer. Properly proportioned, flowing concrete exhibits little
bleeding or segregation and no abnormal segregation. To ensure these properties,
highly angular, flaky or elongated coarse aggregate should be avoided. As far as fine

777
aggregate is concerned, increasing its content by 5 percentage points above the usual
content (with a corresponding reduction in coarse aggregate) contributes to cohesion.
When the fine aggregate is very coarse, an even greater increase in its content may be
required. The reduction in water content has to be taken into account in the
calculation of the yield.
An alternative approach14.6 to ensuring cohesion of flowing concrete is to select the
fine aggregate content so that the total mass of particles smaller than 300 μm in the
aggregate, together with the mass of cementitious material, exceeds 450 kg per cubic
metre of concrete when the maximum size of aggregate is 20 mm; for aggregate
whose maximum size is 40 mm, the ‘ultra-fines’ content should be 400 kg/m3. Instead
of prescribing the content of ‘ultra-fines’ in relation to the cement content of the mix,
their content can be prescribed as a function of the maximum aggregate size. The
Italian Standard for ready-mixed concrete UNI 7163-1979 is cited14.34 as specifying
450 kg (per cubic metre of concrete) of all material smaller than 250 μm when the
maximum size of aggregate is 15 mm; for 20 mm aggregate, the value is 430 kg/m3.
It can be recalled that flowing concrete is well suited to pumping because it offers
less resistance than normal slump concrete so that the rate of pumping can be
increased and pumping over greater distances is possible. Flowing concrete is
desirable for use in large pours because, using a superplasticizer, a low cement
content can be combined with a low water content so that both the heat development
and shrinkage can be kept low. A retarding variety of superplasticizer (Type G
according to ASTM C 494-10) can be beneficial.
Mix selection for high performance concrete
In Table 13.5 details of several high performance concrete mixes were given.
However, a generalized systematic approach to the selection of mix proportions of
high performance concretes has not yet been developed. The reasons for this include
the fact that very few structures made of high performance concrete have, as yet, been
constructed, and each structure involved specific and specially selected materials. For
a future use of high performance concrete, the discussion, in Chapter 13, of the
cement–superplasticizer compatibility and of the influence of the various cementitious
materials, especially silica fume, on the properties of the resultant concrete should be
of considerable value.
Despite the absence of an accepted method of mix selection for high performance
concrete, some specific comments can be made. First, because workability can be
controlled by an appropriate dosage of superplasticizer, the water content should be
chosen on the basis of the water/cement ratio required from strength considerations.
Excessive content of cementitious material should be avoided so as to control
shrinkage: a value of 500 to 550 kg/m3 (850 to 930 lb/yd3), of which about 10 per cent
is silica fume, is a desirable maximum. Portland cement with a high fineness is
preferred. The absolute need for compatibility between Portland cement and the
superplasticizer has already been emphasized. If air entrainment is to be used, the mix
proportions have to be modified by trial and error.14.15
Some help in mix selection for high performance concrete can be obtained from
ACI 211.4R-93,14.16 which is intended to apply to concretes with compressive strengths
(measured on cylinders) between 40 and 80 MPa (or 6000 to 12 000 psi). In this book,
even the latter value is considered to be below what is deemed to be high performance
concrete. Nevertheless, some points are worth noting.

778
 First, with high performance concrete, the specified strength is sometimes required
at ages well beyond 28 days; this should be clearly taken into account in considering
the strength criterion. Second, in some cases, the particular requirement of high
performance concrete is a high modulus of elasticity. To achieve this, the use of
coarse aggregate with a high modulus of elasticity is essential, but it is also important
to select the cementitious materials which lead to a particularly good bond between
the coarse aggregate particles and the matrix.
With respect to the content of coarse aggregate, ACI 211.4R-9314.16 recommends
that the bulk volume of oven-dry rodded coarse aggregate per unit volume of concrete

should be between 0.65 when the maximum aggregate size is 10 mm ( in.) and 0.68
when the maximum aggregate size is 12 mm ( in.) (cf. Table 14.6). It seems that,
unlike ordinary concrete, the value of the bulk volume of coarse aggregate is
unaffected by the fineness modulus of fine aggregate, at least in the range 2.5 to 3.2.
While the broad guidance of ACI 211.4R-9314.16 is useful, it has to be repeated that
an experimental approach to the mix selection to high performance concrete is
unavoidable.
Mix selection for lightweight aggregate concrete
The relation between compressive strength and water/cement ratio applies to concrete
made with lightweight aggregate in the same way as to normal aggregate concrete,
and it is possible to follow the usual procedure of mix selection when lightweight
aggregate is employed. It is very difficult, however, to determine how much of the
total water in the mix is absorbed by the aggregate and how much actually occupies
space within the concrete, that is, forms part of the cement paste. This difficulty is
caused, not only by the very high value of the water absorption of lightweight
aggregates, but also by the fact that the absorption varies widely in rate and, with
some aggregates, may continue at an appreciable rate for several days. A reliable
determination of the specific gravity on a saturated and surface-dry basis is therefore
difficult. This topic is discussed more fully in Chapter 13.
Thus the free water/cement ratio depends on the rate of absorption at the time of
mixing, and not only on the moisture content of the aggregate. Hence, the use of the
water/cement ratio in the calculation of mix proportions is rather difficult. For this
reason, proportioning on the basis of the cement content is prefereable, although, in
the case of rounded lightweight aggregate with a coated or sealed surface and a
relatively low absorption, the use of the standard method of mix selection is
practicable.
Manufactured lightweight aggregate is usually bone-dry, and is rather prone to
segregation. If the aggregate is saturated before mixing, the strength of the resulting
concrete is about 5 to 10 per cent lower than when dry aggregate is used, for the same
cement content and workability. This is due to the fact that, in the latter case, some of
the mixing water is absorbed prior to setting, this water having contributed to the
workability at the time of placing; this behaviour is somewhat similar to that of
vacuum-dewatered concrete. Furthermore, the density of concrete made with a
saturated aggregate is higher, and the resistance of such concrete to freezing and
thawing is impaired. On the other hand, when aggregate with a high absorption is
used, it is difficult to obtain a sufficiently workable and yet cohesive mix, and
generally aggregates with absorption of over 10 per cent should be pre-soaked.

779
 It is interesting to note that initially damp lightweight aggregate usually contains
more total absorbed water after a short immersion in water than initially dry aggregate
immersed for the same length of time. The reason for this is probably that a small
amount of water just moistening an aggregate particle does not remain in the surface
pores but diffuses inward and fills the small pores inside. According to Hanson14.33 this
clears the larger surface pores of water so that, upon immersion, these are open to
ingress of water almost as large as when the aggregate contains no initially absorbed
water.
The preceding discussion explains why mix selection for lightweight aggregate
concrete is best based on the premise that, for a given aggregate, together with a given
air content, and slump, the compressive strength is directly related to the cement
content of the mix. However, this relation can vary widely for lightweight aggregates
from different sources. Figure 14.11 shows examples of this relation for an all-
lightweight aggregate concrete and also for a lightweight aggregate concrete with
normal weight fine aggregate. The practical approach is greatly facilitated by the fact
that, because lightweight aggregate is a manufactured product with properties which
vary very little, the recommendations of the aggregate manufacturer with respect to
mix proportioning make a good starting point in the selection of mix proportions for a
particular purpose.

780
 Fig. 14.11. General relation between compressive strength (measured on
standard cylinders) and cement content for concrete made with: (A) all
lightweight aggregate; (B) normal weight fine aggregate and lightweight coarse
aggregate (based on ref. 14.19)
In the absence of appropriate recommendations or of closely relevant experience,
use can be made of ACI Standard Practice 211.2-91.14.19 The preferred method of ACI
211.2-91 is the so-called volumetric method, which can be used both with all-
lightweight aggregate concrete and with lightweight aggregate concrete containing
normal weight fine aggregate. In this method, the conversion to mass is based on the
damp, loose volume of the aggregates. The total volume of the aggregate is the sum of
the volumes of the separate size fractions: the total loose volume of aggregate, relative

781
to the volume of concrete, is usually between 1.05 and 1.25. Of this total volume of
aggregate, the loose volume of fine aggregate represents between 40 and 60 per cent,
depending on the specific properties of the aggregate used and on the desired
properties of concrete. When the maximum size of aggregate is 20 mm ( in.), it is
convenient to make the first trial mix using equal volumes of fine and coarse
aggregate, and using the cement content corresponding to the desired strength. The
water content used is that which gives the required workability. Because of the
uncertainties involved, it is usual to make three trial mixes, each with a somewhat
different cement content, but all of the required workability. Hence, a relation
between cement content and strength, for the given workability, can be obtained over
a narrow range.
Example
Data similar to those of ACI 211.2-91 will be used. A lightweight aggregate concrete
containing normal weight fine aggregate is required to have a compressive strength
(measured on standard cylinders) of 30 MPa and a maximum air-dry density of 1700
kg/m3. Compliance with the density requirement is determined using ASTM C 567-
05a. The required slump is 100 mm. The damp, loose density of the coarse and the
fine lightweight aggregates is 750 and 880 kg/m3, respectively. The normal weight
fine aggregate has a density in a saturated and surface-dry condition of 1630 kg/m3.
From past experience, for instance as shown in Fig. 14.11, the required cement
content for the trial mix can be taken as 350 kg/m3. The volumes of aggregate to be
used, in cubic metres per cubic metre of concrete, also chosen on the basis of
experience, are: 0.60, 0.19, and 0.34, respectively, for the lightweight coarse,
lightweight fine, and normal weight fine aggregate. Hence, the required quantities for
the first trial batch of 1 m3 are as follows:

The actual density of fresh concrete, which contains some entrapped air, is now
determined using the method of ASTM C 138-09. Supposing the actual density has
been found to be 1660 kg/m3, the yield is 1676/1660 = 1.01. This means that 1 per
cent excess concrete would be produced if the quantities above were used. To remedy
this, all the quantities per cubic metre should be divided by 1.01; for example, the
cement content becomes 350/1.01 = 346 kg/m3.
The density of concrete of 1660 kg/m3 is below the specified maximum and is
close enough to it, but tests to determine the actual strength are necessary.
When adjustments to the mix proportions are necessary, some of the rule-of-thumb
values given in ACI 211.2-9114.19 can be useful. For example, if the mass of fine
aggregate, expressed as a percentage of the total mass of aggregate, is increased by 1
percentage point, the water content necessary to maintain a constant slump should be

782
increased by 2 kg/m3. In order to maintain a constant strength, the cement content
should be increased by about 1 per cent. The mass of coarse aggregate needs to be
reduced so as to maintain the yield.
As another example of the ‘rules’ of ACI 211.2-91, if an increase in slump of 25
mm is required, the water content should be increased by 6 kg/m3. In order to maintain
a constant strength, a concomitant increase in cement content of 3 per cent is
necessary. The mass of fine aggregate needs to be reduced to maintain yield.
Advice on the selection of mix proportions for moderate-strength lightweight
aggregate concrete is given in ACI 523.3R-93.14.20 The same guide gives advice on mix
selection for cellular concrete.
It is worth repeating that the various data on the mix proportions of the lightweight
aggregate concrete are no more than typical figures, different aggregates having
varying values of density and water demand. On the other hand, lightweight aggregate
from a single source possesses a high uniformity. For this reason, mix selection over a
small range of desired properties can be made with considerable confidence.
British method of mix selection (mix design)
The current British method is that of the Department of the Environment revised in
1997.14.11 Similarly to the ACI approach, the British method explicitly recognizes the
durability requirements in the mix selection. The method is applicable to normal
weight concrete made with Portland cement only or also incorporating ground
granulated blastfurnace slag or fly ash, but it does not cover flowing concrete or
pumped concrete; nor does it deal with lightweight aggregate concrete. Three
maximum sizes of aggregate are recognized: 40, 20, and 10 mm.
In essence, the British method consists of 5 steps, as follows.
Step 1. This deals with compressive strength for the purpose of determining the
water/cement ratio. The concept of target mean strength is introduced,
this being equal to the specified characteristic strength plus a margin to
allow for variability. The target mean strength is thus similar in concept
to the mean compressive strength of ACI 318R-0214.8 (see p. 733).
The relation between strength of concrete and the water/cement ratio is dealt with
rather ingeniously. Certain strengths are assumed at a water/cement ratio of 0.5 for
different cements and types of aggregate (Table 14.9). The latter factor recognizes the
significant influence of aggregate on strength. The data of Table 14.9 apply to a
hypothetical concrete of medium richness cured in water at 20 °C (68 °F); richer
mixes would have a relatively higher early strength because they gain strength more
rapidly.

Table 14.9. Approximate Compressive Strengths of Concretes Made with a Free

Water/Cement Ratio of 0.5 According to the 1997 British Method14.11

783
 From Table 14.9, we find the appropriate value of strength (at a water/cement ratio
of 0.5) corresponding to the type of cement, type of aggregate, and age which are to
be used. Turning to Fig. 14.12, we mark a point corresponding to this strength at a
water/cement ratio of 0.5. Through this point, we now draw a curve ‘parallel’ (or,
strictly speaking, affine) to the neighbouring curves. Using this new curve, we read
off (as abscissa) the water/cement ratio corresponding to the specified target mean
strength (as the ordinate). A possible need for a lower water/cement ratio for reasons
of durability must not be forgotten.

784
 Fig. 14.12. Relation between compressive strength and free water/cement ratio
for use in the British mix selection method14.11 (see Table 14.9) (Crown copyright)
Step 2. This deals with the determination of the water content for the required
workability, expressed either as slump or as Vebe time, recognizing the
influence of the maximum size of aggregate and its type, namely
crushed or uncrushed. The relevant data are given in Table 14.10. It can
be noted that the compacting factor is not used in mix selection,
although it can be used for control purposes.

Table 14.10. Approximate Free Water Contents Required to Give Various Levels
of Workability According to the 1997 British Method14.11 (Crown copyright)

785
 Step 3. This determines the cement content, which is simply the water content
divided by the water/cement ratio. This cement content must not conflict
with any minimum value specified for reasons of durability or a
maximum value specified for reasons of heat development.
Step 4. This deals with the determination of the total aggregate content. This
requires an estimate of the fresh density of fully compacted concrete,
which can be read off Fig. 14.13 for the appropriate water content (from
Step 2) and specific gravity of the aggregate. If this is unknown, the
value of 2.6 for uncrushed aggregate and 2.7 for crushed aggregate can
be assumed. The aggregate content is obtained by subtracting from the
fresh density the value of the cement content and of the water content.

786
Fig. 14.13. Estimated wet density for fully compacted concrete14.11 (specific gravity
is given for saturated and surface-dry aggregate) (Crown copyright)
Step 5. This determines the proportion of fine aggregate in the total aggregate,
using the recommended values of Fig. 14.14; only data for 20 and 40
mm aggregates are shown. The governing factors are: the maximum size
of aggregate, the level of workability, the water/cement ratio, and the
percentage of fine aggregate passing the 600 μm sieve. Other aspects of
the grading of the fine aggregate are ignored and so is the grading of the
coarse aggregate. Once the proportion of fine aggregate has been
obtained, multiplying it by the total aggregate content gives the content
of fine aggregate.

787
788
Fig. 14.14. Recommended proportion of fine aggregate (expressed as percentage
of total aggregate) as a function of free water/cement ratio for various
workabilities and maximum sizes14.11 (numbers refer to percentage of fine
aggregate passing 600 μm sieve) (Building Research Establishment; Crown
copyright)

789
 The content of coarse aggregate is then the difference between the total aggregate
content and the content of fine aggregate. The coarse aggregate, in turn, should be
divided into size fractions depending on the aggregate shape. As a general guide, the
percentages of Table 14.11 can be used.

Table 14.11. Proportion of Coarse Aggregate Fractions According to the 1997

British Method14.11

Following the above calculations, trial mixes must be made. It should also be
remembered that the British method is based on the experience of British materials so
that the various values given in the tables and figures may not be applicable in other
parts of the world.
The selection of mix proportions to achieve a desired splitting tension strength,
which was previously included in the British method, is no longer
recommended. More generally, in British practice, although flexural strength may be
the correct design criterion for some structures, e.g. highway pavements, the selection
of mix proportions on the basis of a direct determination of the flexural strength is
rarely practised. The reason for this lies in the difficulty of the use of the modulus of
rupture as a control test (see p. 599). Thus, mix proportions are selected in the usual
manner, and both the compressive and tensile strengths are determined. Provided the
latter is adequate, control and the attendant mix adjustments are based on the
compressive strength.
The British method of mix selection may become modified when the European
Standard BS EN 206-1 : 2000 becomes widely used. Furthermore, the European
Standards have not yet ‘shaken down’ and are likely to be amended.
Example
We wish to select a mix to satisfy requirements similar to those used in the example
of the American method of mix selection (p. 757). These are: a mean 28-day
compressive strength (measured on standard cubes) of 44 MPa (which is equivalent to
a cylinder strength of 35 MPa); a slump of 50 mm; uncrushed aggregate with a
maximum size of 20 mm; specific gravity of aggregate of 2.64; 60 per cent of fine
aggregate passes the 600 μm sieve; no air entrainment required; ordinary Portland
cement to be used.
Step 1. From Table 14.9, for ordinary Portland cement and uncrushed aggregate,
we find the 28-day strength to be 42 MPa. We enter this value on the
ordinate corresponding to a water/cement ratio of 0.5 in Fig. 14.12; this
point is marked A. Through A, we draw a line ‘parallel’ to the nearest
curve until it intersects the ordinate corresponding to the specified
strength of 44 MPa; this is point B. The ordinate through this point gives
the water/cement ratio of 0.48.
Step 2. From Table 14.10, for 20 mm uncrushed aggregate and a slump of 50
mm, we find the water requirement to be 180 kg/m3.

790
 Step 3. The cement content is 180/0.48 = 375 kg/m3.
Step 4. From Fig. 14.13, for a water content of 180 kg/m3 and aggregate with a
specific gravity of 2.64, we read off the fresh density of concrete of
2400 kg/m3. The total aggregate content is thus:

2400 – 375 – 180 = 1845 kg/m3.

Step 5. In Fig. 14.14, we find the particular diagram for the maximum size of
aggregate of 20 mm and a slump encompassing the value of 50 mm. On
the line representing fine aggregate with 60 per cent passing the 600 μm
sieve, at a water/cement ratio of 0.48, the proportion of fine aggregate is
32 per cent (by mass of total aggregate). Hence, the fine aggregate
content is:

32 × 1845 = 590 kg/m3

and the coarse aggregate content is 1845 – 590 = 1255 kg/m3.

Other methods of mix selection
It is not suggested that, on each occasion, the selection of mix proportions should
necessarily follow any of the procedures described earlier. Indeed, various people
have their own methods which work well. What these ‘methods’ have in common is
that they use shortcuts or rule-of-thumb steps in the procedure which are based on an
individual’s experience. As long as these ‘methods’ are used by the same individual
and as long as the materials involved are not fundamentally different from those used
in the past, all is well. If, however, a person has to select mix proportions using
unfamiliar materials, the procedures described in this chapter are very helpful. But,
even so, selecting mix proportions is not just a rule-based process.
Over the years, numerous attempts have been made to develop mix-proportioning
equations based on observed influences of various factors. Such relations, or models,
inevitably represent averages of behaviour. And yet, in every particular case, the
behaviour of concrete is affected by properties of ingredients which cannot be, or
cannot yet be, expressed mathematically. These properties include aggregate shape
and texture which, at present, are described only in broad terms such as ‘angular
shape’ or ‘smooth texture’. Likewise, grading of aggregate is measured only at several
sieve apertures, between any two of which there may be variation in the actual size of
the particles. There is little prospect of a proper quantification of these properties in
the near future. The possibility of determining these properties of aggregate during
batching, so that the quantity of water added can be instantly adjusted, is even more
remote.
Many properties of cement are also not properly included in the various models
because the actual properties of the cement used in a given mix (as distinct from
average properties) are not known or not determined.
These average relations may be valid ‘on average’, but trying to use them with a
particular set of materials must perforce be subject to large errors. It is, therefore,
futile to use elegant computer-based calculations of mix proportions. This is not to say
that such an approach may not be feasible in the future when it becomes possible to
describe mathematically the properties of all the materials to be used, and also to
control, or measure, these properties at the batcher.14.35

791
 One other note of caution may be appropriate. A statistically derived model can, at
best, be valid within the range of variables used in deriving it. If this range is not
clearly stated, unwitting extrapolation may be strongly misleading. It is also worth
adding that some of the more elaborate methods involve numerous interacting terms,
but there is little value in including factors which are subject to unpredictable
variation during construction. Thus, the promise of a ‘hands-off’ computer-
controlled definitive selection of mix proportions is unrealistic. In the meantime, the
selection of mix proportions must be based on preliminary calculations of the kind
described in this chapter, followed by trial mixes. The selection of mix proportions is
an art as much as a science.
Concluding remarks
The various methods of mix selection may seem simple and, indeed, they do not
involve any complex calculations. However, a successful implementation of the
selection requires experience, coupled with the knowledge of the influence of various
factors upon the properties of concrete; this knowledge must be based on
an understanding of the behaviour of concrete. When these three desiderata –
experience, knowledge, and understanding – are all present, the first trial mix is likely
to be approximately satisfactory, and can be rapidly and successfully adjusted so as to
achieve a mix with the desired properties.
It is not enough to select a suitable concrete mix; it is also necessary to ensure a
proper execution of all the operations involved in concreting. Such execution requires
skill backed by appropriate knowledge at the execution level. The belief, once held,
that any fool can make concrete has, alas, sometimes led to a situation where he did.
The consequences of such execution manifest themselves before long. It cannot be
stated too strongly that, competently used, concrete is a very successful construction
material but, in the literal sense of the word, concrete is not foolproof.
The first edition, and the subsequent two editions of this book, ended with a
‘tongue-in-cheek’ note saying, “If the reader is unable to design a satisfactory mix he
should seriously consider the alternative of construction in steel.” The situation has
changed. First, the reader may as well be a she as a he. Then, for many modern
structures, steel is not a simple alternative and may not be appropriate. And, lastly, in
this fifth and truly final edition well into the third millennium, perhaps such a
ubiquitous and weighty material as concrete should not be treated too flippantly. The
aim of this book has been to try to provide an understanding of the behaviour of
concrete – an excellent construction material for many years to come. If this aim has
been achieved, the reader will not, in despair and frustration, need to “seriously
consider the alternative of construction in steel”.
References

14.1. G. M. CAMPBELL and R. J. DETWILER, Development of mix designs for strength

and durability of steam-cured concrete, Concrete International, 15, No. 7, pp.
37–9 (1993).
14.2. W. C. GREER, JR, Variation of laboratory concrete flexural strength
tests, Cement, Concrete and Aggregates, 5, No. 2, pp. 111–22 (Winter 1983).
14.3. D. S. LANE, Flexural strength data summary, NRMCA Technical Information
Letter, No. 451, 5 pp. (Silver Spring, Maryland, 1987).

792
 14.4. ACI 211.3-75, Revised 1987, Reapproved 1992, Standard practice for
selecting proportions for no-slump concrete, ACI Manual of Concrete
Practice, Part 1: Materials and General Properties of Concrete, 11 pp.
(Detroit, Michigan, 1994).
14.5. ACI 211.1-91, Standard practice for selecting proportions for normal,
heavyweight, and mass concrete, ACI Manual of Concrete Practice, Part 1:
Materials and General Properties of Concrete, 38 pp. (Detroit, Michigan,
1994).
14.6. P. C. HEWLETT, Superplasticised concrete: Part 1, Concrete, 18, No. 4, pp. 31–2
(London, 1984).
14.7. ACI 1 1R-85, Quality assurance systems for concrete construction, ACI
Manual of Concrete Practice, Part 2: Construction Practices and Inspection
Pavements, 7 pp. (Detroit, Michigan, 1994).
14.8. ACI 318-02, Building code requirements for structural concrete, ACI Manual
of Concrete Practice, Part 3: Use of Concrete in Buildings – Design,
Specifications, and Related Topics, 443 pp.
14.9. J. KRELL and G. WISCHERS, The influence of fines in concrete on consistency,
strength and durability, Beton, 38, No. 9, pp. 356–9 and No. 10, pp. 401–4
(1988) (British Cement Association translation).
14.10. CP 114 : 1969, The Structural Use of Reinforced Concrete in Buildings,
British Standards Institution, 94 pp. (London, 1969).
14.11. D. C. TEYCHENNE, R. E. FRANKLIN AND H. C. ERNTROY, Design of Normal
Concrete Mixes, 42 pp. (Watford, U.K., Building Research Establishment,
1997).
14.12. CP 114 (1948), The Structural Use of Reinforced Concrete in Buildings,
British Standards Institution, 54 pp. (London, 1948).
14.13. R. D. GAYNOR, Ready mixed concrete, in Concrete and Concrete-making
Materials, Eds P. Klieger and J. F. Lamond, ASTM Sp. Tech. Publ. No. 169C,
pp. 511–21 (Philadelphia, Pa, 1994).
14.14. STRATEGIC HIGHWAY RESEARCH PROGRAM, SHRP-C/FR-91-103, High Performance
Concretes: A State-of-the-Art Report, 233 pp. (Washington DC, NRC, 1991).
14.15. M. LESSARD et al., Formulation d’un béton à hautes performances à air
entrainé, Bulletin Liaison Laboratoires des Ponts et Chaussées, 189, pp. 41–
51 (Nov.–Dec. 1993).
14.16. ACI 221.4R-93, Guide for selecting proportions for high-strength concrete
with portland cement and fly ash, ACI Manual of Concrete Practice, Part 1:
Materials and General Properties of Concrete, 13 pp. (Detroit, Michigan,
1994).
14.17. ACI 225.R-92, Guide to the selection and use of hydraulic cements, ACI
Manual of Concrete Practice, Part 1: Materials and General Properties of
Concrete, 29 pp. (Detroit, Michigan, 1994).
14.18. ACI 214-77 (Reapproved 1989), Recommended practice for evaluation of
strength test results of concrete, ACI Manual of Concrete Practice, Part 2:
Construction Practices and Inspection Pavements, 14 pp. (Detroit, Michigan,
1994).

793
 14.19. ACI 211.2-92, Standard practice for selecting proportions for structural
lightweight concrete, ACI Manual of Concrete Practice, Part 1: Materials
and General Properties of Concrete, 14 pp. (Detroit, Michigan, 1994).
14.20. ACI 523.3R-93, Guide for cellular concretes above 50 pfc, and for aggregate
concretes above 50 pfc with compressive strengths less than 2500 psi, ACI
Manual of Concrete Practice, Part 5: Masonry, Precast Concrete, Special
Purposes, 16 pp. (Detroit, Michigan, 1994).
14.21. R. HEGNER, Les résistances du béton selon la norme SIA 162 (1989), Bulletin
du Ciment, 57, No. 21, 12 pp. (Wildegg, Switzerland, 1989).
14.22. ACI 363R-92, State-of-the-art report on high-strength concrete, Manual of
Concrete Practice, Part 1: Materials and General Properties of Concrete,
55 pp. (Detroit, Michigan, 1994).
14.23. D. P. MCNICHOLL and B. WONG, Investigation appraisal and repair of large
reinforced concrete buildings in Hong Kong, in Deterioration and Repair of
Reinforced Concrete in Arabian Gulf, Vol. 1, Bahrain Society of Engineers,
pp. 327–40 (Bahrain, 1987).
14.24. J. E. COOK, 10,000 psi concrete, Concrete International, 11, No. 10, pp. 67–75
(1989).
14.25. P. N. BALAGURU and V. RAMAKRISHNAN, Authors’ closure to paper in ACI
Materials Journal, 84, No. 1, 1987, ACI Materials Journal, 85, No. 1, p. 60
(1988).
14.26. A. M. NEVILLE, The relation between standard deviation and mean strength of
concrete test cubes, Mag. Concr. Res., 11, No. 32, pp. 75–84 (July 1959).
14.27. H. C. ERNTROY, The variation of works test cubes, Cement Concr. Assoc.
Research Report No. 10 (London, Nov. 1960).
14.28. H. RÜSCH, Zur statistischen Qualitätskontrolle des Betons (On the statistical
quality control of concrete), , 6, No. 11, pp. 387–94
(1964).
14.29. ACI COMMITTEE 214, Recommended practice for evaluation of strength test
results of concrete, (ACI 214-77), and Commentary, J. Amer. Concr.
Inst., 73, No. 5, pp. 265–78 (1976).
14.30. H. H. NEWLON, Variability of portland cement concrete, Proceedings, National
Conf. on Statistical Quality Control Methodology in Highway and Airfield
Construction, pp. 259–84 (Univ. of Virginia School of General Studies,
Charlottesville, 1966).
14.31. J. B. KENNEDY and A. M. NEVILLE, Basic Statistical Methods for Engineers and
Scientists, 3rd Edn, 613 pp. (New York and London, Harper and Row, 1986).
14.32. N. PETERSONS, Ready mixed concrete in Sweden, CBI Reports 5:77, 15 pp.
(Swedish Cement and Concrete Research Inst., 1977).
14.33. J. A. HANSON, American practice in proportioning lightweight-aggregate
concrete, Proc. 1st Int. Congress on Lightweight Concrete, Vol. 1: Papers,
pp. 39–54 (London, Cement and Concrete Assoc., May 1968).
14.34. M. A. ALI, A Review of Italian Concreting Practice, Building Research
Establishment Occasional Paper, 25 pp. (July 1992).

794
 14.35. A. M. NEVILLE, Is our research likely to improve concrete?, Concrete
International, 17, No. 3, pp. 45–7 (1995).

795